• <tr id="yyy80"></tr>
  • <sup id="yyy80"></sup>
  • <tfoot id="yyy80"><noscript id="yyy80"></noscript></tfoot>
  • 99热精品在线国产_美女午夜性视频免费_国产精品国产高清国产av_av欧美777_自拍偷自拍亚洲精品老妇_亚洲熟女精品中文字幕_www日本黄色视频网_国产精品野战在线观看 ?

    基于無人機圖像顏色與紋理特征的小麥不同生育時期生物量估算

    2022-05-30 22:30:33戴冕楊天樂姚照勝劉濤孫成明
    智慧農業(yè)(中英文) 2022年1期
    關鍵詞:生物量小麥

    戴冕 楊天樂 姚照勝 劉濤 孫成明

    摘要:為實現(xiàn)小麥生物量田間快速無損監(jiān)測,開展基于不同密度、氮肥和品種處理的田間試驗,應用無人機獲取小麥越冬前期、拔節(jié)期、孕穗期和開花期4個時期的 RGB 圖像,通過影像處理獲取小麥顏色指數(shù)和紋理特征參數(shù),并同時期通過田間取樣獲取小麥生物量;分析不同顏色指數(shù)和紋理特征參數(shù)與小麥生物量的關系,篩選出適合小麥生物量估算的顏色和紋理特征指數(shù)。結果表明,不同時期圖像顏色指數(shù)和小麥生物量均有較高的相關性,且大部分達到極顯著相關水平;圖像紋理特征指數(shù)與小麥生物量的相關性較差,只有少數(shù)指標達到顯著或極顯著相關水平?;谏鲜鼋Y果,研究利用相關性最高的顏色指數(shù)或顏色指數(shù)與紋理特征指數(shù)結合構建小麥不同生育時期的生物量估算模型,并通過獨立的實測生物量數(shù)據(jù)對模型進行了驗證,模型模擬值與實測值之間的相關性均達到了極顯著水平( P<0.01), RMSE 均較小。其中,顏色指數(shù)模型在4個時期的 R2分別為0.538、0.631、0.708和0.464,RMSE 分別為27.88、516.99、868.26和1539.81 kg/ha。而顏色和紋理指數(shù)結合的模型在4個時期的 R2分別為0.571、0.658、0.753和0.515, RMSE 分別為25.49、443.20、816.25和1396.97 kg/ha ,說明模型估算的結果是可靠的,且精度較高。同時結合無人機圖像顏色和紋理特征指數(shù)的小麥生物量估測模型的效果要優(yōu)于單一顏色指數(shù)模型。研究可為小麥田間長勢實時監(jiān)測與生物量估算提供新的手段。

    關鍵詞:小麥;無人機圖像;顏色指數(shù);紋理特征指數(shù);生物量;紋理指數(shù)

    Wheat Biomass Estimation in Different Growth Stages Based on Color and Texture Features of UAV Images

    DAI Mian1,2 , YANG Tianle1,2 , YAO Zhaosheng1,2 , LIU Tao1,2 , SUN Chengming1,2*

    (1. Jiangsu Key Laboratory of Crop Genetics and Physiology/Jiangsu Key Laboratory of Crop Cultivation andPhysiology, Agricultural College, Yangzhou University, Yangzhou 225009, China;2. Jiangsu Co-Innovation Centerfor Modern Production Technology of Grain Crops, Yangzhou University, Yangzhou 225009, China )

    Abstract: In order to realize the rapid and non-destructive monitoring of wheat biomass, field wheat trials were con‐ ducted based on different densities, nitrogen fertilizers and varieties, and unmanned aerial vehicle (UAV) was used to obtain RGB images in the pre-wintering stage, jointing stage, booting stage and flowering stage of wheat. The col‐ or and texture feature indices of wheat were obtained using image processing, and wheat biomass was obtained by manual field sampling in the same period. Then the relationship between different color and texture feature indices and wheat biomass was? analyzed to? select the? suitable? feature? index? for wheat biomass? estimation. The results showed that there was a high correlation between image color index and wheat biomass in different stages, the val‐ues of r were between 0.463 and 0.911(P<0.05). However, the correlation between image texture feature index and wheat biomass was poor, only 5 index values reached significant or extremely significant correlation level. Based on the above results, the color indices with the highest correlation to wheat biomass or the combining indices of color and texture features in different growth stages were used to construct estimation model of wheat biomass. The mod‐els were validated using independently measured biomass data, and the correlation between simulated and measured values reached the extremely significant level (P<0.01), and root mean square error (RMSE) was smaller. The R2 of color index model in the four stages were 0.538, 0.631, 0.708 and 0.464, and RMSE were 27.88, 516.99, 868.26 and 1539.81 kg/ha, respectively. The R2 of the model combined with color and texture index were 0.571, 0.658, 0.753 and 0.515, and RMSE were 25.49, 443.20, 816.25 and 1396.97 kg/ha, respectively. This indicated that the estimated results using the models were reliable and accurate. It also showed that the estimation models of wheat biomass com ‐bined with color and texture feature indices of UAV images were better than the single color index models.??????? Key words: wheat; UAV? image; color? index; texture? feature? index; biomass; texture index

    CLC number:S512???????????????????? Documents code:A????????????????? Article ID:SA202202004

    Citation:DAI Mian, YANG Tianle, YAO Zhaosheng, LIU Tao, SUN Chengming. Wheat biomass estimation in dif‐ferent growth stages based on color and texture features of UAV images[J]. Smart Agriculture, 2022, 4(1):71-83.(in English with Chinese abstract)

    戴冕, 楊天樂, 姚照勝, 劉濤, 孫成明.基于無人機圖像顏色與紋理特征的小麥不同生育時期生物量估算[J].智慧農業(yè)(中英文), 2022, 4(1):71-83.

    1? Introduction

    Biomass is an important physical and chemical parameter in ecosystems and a significant index for assessing? the? life? activities? of vegetation? and? for monitoring growth and estimating crop yield[1]. Tra‐ditional? biomass? estimation? methods? are? not? only time-consuming? and? labor-intensive,? but? also? can not conduct large-scale monitoring[2]. The rapid de‐velopment? of remote? sensing technology? in recent years? has? accelerated? its? wide? application? in? crop biomass estimation, as it is fast, accurate, and non- destructive[1].

    Based? on? the? spectral? features? of vegetation, previous studies have obtained many achievements in using vegetation indices to estimate biomass. Ji‐menez-Sierra? et? al.[3]? proposed? GBF-Sm-Bs? ap ‐proach by obtaining biomass estimation correlation of 0.995 with R2 =0.991. This result increased the precision? in? the? biomass? estimation? by? about 62.43% compared? to? previous? work. Hou? et? al.[4] built a biomass estimation model based on multiple vegetation indices, of which the index model of red edge location was used to estimate the wheat bio ‐ mass model, R2 were all greater than 0.8. Liu et al.[5] used arbor as an object to analyze the correlation be‐ tween biomass and biomass factors, and constructed a highly? accurate? forest biomass estimation model using the partial least-squares method, the correla‐tion coefficient between the predicted value and the measured? value? was 0.718,? and? the? accuracy? was 90.1%. However, the vegetation index is not sensi‐tive to canopy biomass changes in high-density sce‐narios and deviates easily when inverting physical and chemical parameters or even agronomic param ‐eters, thus affecting the accuracy of the estimation model[6]. In? response,? some? scientists? have? added texture? feature information to the? spectral features to improve? saturation when only? spectral informa‐tion is available, as well as to improve the spatialand? temporal? discrimination? of image? informationand the estimation potential of agronomic parame‐ters[7].

    Some researchers have quantitatively analyzedthe application of texture features in agronomic pa‐rameters. Gu? et? al.[8]? studied? vegetation? coverageand found that the combination of vegetation indi‐ces and texture feature indices could improve the ac‐curacy? of? vegetation? coverage? estimation. Sarkerand Nichol[9]? assessed? the? biomass? of arbor? forestand found that the combination of vegetation indi‐ces and texture features was better at biomass esti‐mation. Cao et al.[10] extracted the texture featuresand? spectral? features? of thematic? mapper? imagesand? established? a? biomass? regression? estimationmodel,? which? could? effectively? estimate? the? bio ‐mass of mangrove wetlands. Mu et al.[11] constructeda multivariate regression model of vegetation indi‐ces, texture features, and vegetation biomass usingfour typical vegetation indices, and R2 of the modelwas 0.713, the minimum RMSE was 98.2543 g/m2.Based on the above researches, it was found that theaccuracy? of the? model? for? estimating? biomass? incombination with texture? features was higher thanthe? accuracy? of the model? for? estimating biomassusing? a? single? vegetation? index? with? color? fea‐tures[12].

    In terms of monitoring crop growth based onunmanned aerial vehicle (UAV) images, Lee et al.[13]used RGB (Red Green Blue) cameras to obtain ricecanopy images and found that the absolute value ofgreen light depth (G) simulated by the exponentialequation was well correlated with the abovegroundbiomass and leaf area index (LAI). Wang? et al.[14]demonstrated that the G-R value of rice RGB imag‐es had a good relationship with biomass and LAI,which could be used to construct an estimation mod‐el. Wang et al.[15] used random forest-support vector regression (RF-SVR) model? to? estimate? soil? and plant? analyzer? development (SPAD) values? of the wheat? canopy? and? achieved? high? accuracy (R2=0.754, RMSE=1.716). Yue et al.[16] combined UAV remote? sensing? and AquaCrop? to? estimate? winter- wheat biomass, the predicted biomass agreed with the? measured? values (R2=0.61,? RMSE=2.10 t/ha). Shan et al.[17] used explored the vertical distribution of wheat biomass and found that a linear regression relationship could be better obtained with a single row of wheat images, and the estimation accuracy was also higher than multi-row wheat image.

    In this study, UAVs were used to obtain RGB images of wheat in different growth periods, and si‐multaneous sampling was conducted in the field to determine wheat biomass. The best color and tex‐ture feature indices were determined through corre‐lation analysis, and these indices were used to con ‐ struct? the? estimation? models? of wheat? biomass? in different growth periods based on UAV platform in order to provide a new method for biomass estima‐tion? and? real-time? monitoring? of wheat? growth? in the field.

    2? Materials and methods

    2.1 Field experiment design

    Two experiments were conducted in this study, of which the data from Experiment 1 were used to construct a wheat biomass estimation model and the data from Experiment 2 were used for model valida‐tion. Study? sites were located in Yizheng city andZhangjiagang city, Jiangsu province.

    2.1.1? Experiment 1 for estimation model???? Experiment 1 was conducted in Yizheng from

    2018 to 2019. Two? wheat varieties? of Yangmai23 and Yangfumai4 were? selected as the research ob‐jects. Three planting densities were respectively setas:1 million plants/ha, 1.5 million plants/ha, and 2million plants/ha. Four nitrogen fertilizer levels wererespectively set as:0 kg/ha, 120 kg/ha, 160 kg/ha,and 200 kg/ha. Nitrogen fertilizers were applied ac‐cording? to? the? ratio? of base? fertilizer: new? shootsboosting fertilizer: jointing fertilizer: booting fertil‐izer =5:1:2:2, and the phosphorus and potassiumfertilizers? were? applied? according? to? the? ratio? ofbase fertilizer: jointing fertilizer =5:5, and the appli‐cation? amounts? all? were 120 kg/ha. Wheat? wasplanted on November 2, 2016, with a plot area of16.65 m2. Each treatment was repeated twice for atotal of 48 experimental plots.

    2.1.2? Experiment 2 for model validation

    Experiment 2 was conducted in Zhangjiagangfrom 2017 to 2018. The wheat variety, density, andfertilizer? were? the? same? as? Experiment 1. Wheatwas planted on November 10, 2017, with a plot areaof 30 m2. Each treatment was repeated twice for atotal of 48 experimental plots.

    2.2 Data acquisition method

    2.2.1? Image acquisition device

    The Inspire 1 RAW UAV (DJI, Shenzhen, Chi‐na) was used for image data acquisition. This UAVis small, powerful, convenient and easy to operate,equipped with a 16-megapixel camera and is able tofly for 15-20 min depending on the load. The re‐mote control was connected to the wireless followerto extend the control distance to 5 km.

    2.2.2? Image acquisition

    UAV image? acquisition runs through the pre-wintering? period, jointing? period,? booting? period,and flowering period, and sampling time was about8:00-10:00 AM or 3:00-5:00 PM. In this study, DJIGS? Pro? was? used? to? automatically? generate? flightroutes? within? the? designated? area,? and? automaticflight,? automatic? shooting,? and? complete? relevantdata receiving, processing? and? sending. The? flight route of UAV was s-type, and the flight altitude was set to 15 m. In order to achieve accurate image reg ‐istration, the image repetition rate was set to 60% on the main route and 70% between the main route during route and point planning. The standard cali‐bration panels were used to perform radiation cali‐bration on RGB band sensors during flight to mini‐mize the influence of constantly changing lighting conditions on RGB images. After the aerial images were collected, the orthophoto image was generated by using the software Pix4DMapper, and the imag‐ es were stitched together seamlessly through feature matching of adjacent images[18].

    2.2.3? Determination of aboveground biomass

    In? the? pre-wintering? period,? jointing? period, booting? period,? and? flowering? period? of? wheat growth, 15 wheat? plants? were? selected? from? each plot, the aboveground parts of plants were collected and? transported? back? to? the? laboratory,? and? were dried in? an oven? for 1.5 h? at 105°C? and then? at 80°C until constant weight, and then were weighed and converted to biomass per unit area.

    2.3 Data analysis and utilization

    2.3.1? UAV image preprocessing method?? MATLAB2014a? software? was? used? for? UAVimage preprocessing, including image cropping, de‐noising,? smoothing,? and? sharpening. Image? crop ‐ping? stitches? the? images? into? uniform? images? ac‐cording to different cells. Denoising eliminates thenoise in digital images. Smoothing and sharpeningreduce the? slope of the image, improve the imagequality, and reduce the loss of pixel extraction fromthe target.

    2.3.2 Color indices

    Eight? commonly? used? color? indices? were? se‐lected for UAV image data analysis, including visualatmospheric resistance vegetation index (VARI)[19] ,excess red vegetation index (ExR)[20] , excess greenvegetation index (ExG)[21] , green leaf vegetation in ‐dex (GLI)[22] ,? excess? green-red? difference? index(ExGR)[20] ,? normalized? difference? index (NDI)[23] ,modified? green? red? vegetation? index (MGRVI)[24]and red green blue vegetation index (RGBVI)[24] , asshown in Table 1.

    2.3.3 Texture feature indices

    MATLAB software was used to extract texture features? based? on? a? gray? level? co-occurrence? ma‐trix[25]. Four common texture features were extractedfrom? the? UAV? image: angular? second? moment(ASM), contrast (CON), correlation (COR) and en‐tropy (ENT)[25] , as shown in Table 2.

    2.4 Model construction and evaluation

    Based on the correlation between image color and texture feature indices and wheat biomass, the color? and? texture? feature? indices? with? the? largest correlation coefficient were selected to construct the regression? model? for? biomass? estimation (Experi‐ment 1 data). Then the independently measured bio ‐ mass? data? were? used? to? validate? and? evaluate? the model? based? on? the? coefficient? of? determination(R2), root mean square error (RMSE), and 1:1 map(Experiment 2 data).

    3? Results and analysis

    3.1 Correlation? between? wheat? biomass and? image color/texture feature? indi ? ces in different growth periods

    Wheat? biomass? experiences? various? changesthroughout the growth period and undergoes a pro ‐cess of continuous increase. In this study, the datain Experiment 1 were used to quantitatively analyzethe correlation between eight color indices as wellas four texture feature indices in the main growthperiods of wheat and biomass to determine the opti‐mal color index and texture feature index for esti‐mating biomass. The correlation between the differ‐ent color indices as well as the texture feature indi‐ces and wheat biomass based on the UAV image areshown in Table 3 and Table 4.

    3.2 Biomass?? estimation?? models?? for wheat growth in different growth peri ? ods based on color indices

    3.2.1? Model construction

    (1) Estimation model of wheat biomass in thepre-wintering period

    It can be seen from Table 1 that the correlationbetween? the? color? indices? of the? UAV? image? andbiomass was good during this period. Besides NDI,the correlation between the other seven indices andbiomass was extremely significant (P<0.01), among which the correlation of VARI and biomass was the highest, with the correlation coefficient r reaching 0.743. Therefore, the color index VARI was chosen as the independent variable. The wheat biomass esti‐mation model was constructed by using regression analysis:

    B1=535.9×VARI+87.9?? ????????????(13)

    where B1 represents the biomass of wheat in the pre- winter period, kg/ha. The coefficient of determina‐tion R2 was 0.553.

    (2) Estimation model of wheat biomass in the jointing period

    The correlation between the color indices andbiomass? of the UAV? image? in? the jointing period was the highest, and eight color indices reached a highly? significant? correlation? level. Among? them, ExGR had the highest correlation with biomass, and the? correlation? coefficient? r reached 0.911. There‐ fore, the color index ExGR was selected as the inde‐ pendent variable, and the wheat biomass estimation model is:

    B2=9054.6×ExGR+1915.5??????????? (14)

    where? B2? represents? the? biomass? of wheat? in? the jointing period, kg/ha. It was constructed using re‐gression? analysis, with? a? coefficient? of determina‐tion R2 of 0.804.

    (3) Estimation model of wheat biomass in the booting period

    With? the? gradual? advancement? of? wheat growth and the increasing biomass, the color of the UAV image will become saturated, and its correla‐tion? with? biomass? will? be? correspondingly? weak‐ened. The correlation between UAV image color in ‐ dices and biomass in the booting period was signifi‐cantly lower than that in the jointing period. Only six? indices? were? significantly? correlated. Among them, MGRVI had the highest correlation with bio ‐mass,? and the? correlation? coefficient? r was 0.817.Therefore, the color index MGRVI was chosen? asthe independent variable, and the wheat biomass es‐timation model is:

    B3=20024.1×MGRVI?543.2????????? ( 15)

    where? B3? represents? the? biomass? of wheat? in? thebooting period, kg/ha. It was constructed using re‐gression analysis, with an R2 of 0.670.

    (4) Estimation model of wheat biomass in theflowering period

    The canopy image in the flowering period con ‐tains different types of objects such as the ear andleaf, and the correlation between image color indi‐ces and biomass is further reduced. During this peri‐od, the correlation between the UAV image color in ‐dices and biomass was the lowest among the fourperiods, three? of which being non-significant,? onebeing? significantly? correlated,? and? the? other? fourreaching extremely significant levels, of which thecorrelation? of VARI? and? biomass? was? the? highestand the correlation coefficient r was 0.679. There‐fore, the color index VARI was selected as the inde‐pendent variable, and the wheat biomass estimationmodel is:

    B4=42623.1×VARI+7115.3?????????? (16)

    where? B4? represents? the? biomass? of wheat? in? theflowering period, kg/ha. It was constructed using re‐gression analysis, and the R2 was 0.461.

    3.2.2? Model validation

    (1) Validation of the wheat biomass estimationmodel in the pre-wintering period

    Independently measured data were used to vali‐date the wheat biomass estimation model in the pre-wintering period? and plot the 1:1 relationship be‐tween the measured values? and the model predic‐tions (Fig.1(a)). It is evident from the figure thatthere? was? high? agreement? between? the? predictedand measured values of wheat biomass in the pre-wintering period, and the predicted R2 of the modelwas 0.538. Correlation analysis showed that the cor‐ relation between the predicted value and the mea‐sured value reached a highly significant level, indi‐cating that the estimation model is feasible. In addi‐tion, the simulated RMSE was 27.88 kg/ha, which is relatively small, indicating that the simulation re‐sults of the model are relatively reliable.

    (2) Validation of the wheat biomass estimation model in the jointing period

    The validation results of the estimation model in the jointing period are shown in Figure 1(b). The predicted value of wheat biomass in the jointing pe‐riod was close to the measured value, and there was high consistency between the two. The predicted R2 of the model was 0.631, and the effect was better than the pre-wintering period. Correlation? analysis indicated that the correlation between the predicted value and the measured value reached a highly sig ‐nificant level, indicating that the estimation model is? feasible. In? addition,? the? simulated? RMSE? was 516.99 kg/ha, which is relatively? small, indicating that the simulation results of the model are relative‐ly reliable.

    (3) Validation of the wheat biomass estimation model in the booting period

    The results of the estimation model in the boot‐ing period are shown in Figure 1(c). It can be seen from? the? figure? that? the? predicted? value? of wheat biomass in the booting period was close to the mea‐sured? value,? and? there? was? high? consistency? be‐ tween the two. The predicted R2 of the model was 0.708, and the effect was better than the jointing pe‐riod. Correlation analysis indicated that the correla‐tion between the predicted value and the measured value was significant, indicating that the estimation model is feasible. In addition, the simulated RMSE was 868.26 kg/ha, which is relatively small, indicat‐ing that the simulation results of the model are rela‐tively reliable.

    (4) Validation of the wheat biomass estimationmodel in the flowering period

    The validation results of the estimation modelin the? flowering period? are? shown? in Figure 1(d).The predicted value of wheat biomass in the flower‐ing? period? was? close? to? the? measured? value,? andthere was a good agreement between the two. Themodel predicted an R2 of 0.464, and the effect wasthe poorest? among the? four periods. However, thecorrelation analysis demonstrated a correlation be‐tween the predicted value and the measured value,indicating that the estimation model is feasible. Inaddition, the simulated RMSE was 1539.81 kg/ha,which is relatively small, indicating that the simula‐tion results of the model are relatively reliable.

    In? summary,? the? UAV? image? color? indicescould be used to estimate wheat biomass in differ‐ent growth periods, and the effects in the differentperiods also differed. However, wheat biomass esti‐mation using? a? single? color index? did not achievethe best results.

    3.3 Wheat biomass estimation models in different? growth? periods? based? on color and texture feature indices

    3.3.1? Model construction

    (1) Estimation model of wheat biomass in thepre-wintering period

    It can be seen from Table 4 that the correlationbetween UAV image texture indices and biomass inthe pre-wintering period was high, however, it waspoorer than the color indices. Among the four tex‐ture feature indices, two were not significantly cor‐related,? whereas? the? other? two? were? significantlycorrelated, of which ASM had the highest correla‐tion? with? biomass,? with? a? correlation? coefficientof 0.573. Therefore,? the? texture? feature? index

    ASM was combined with the color index VARI, andthe? wheat? biomass? estimation? multiple? regressionmodel is:

    B5=547.9×VARI?35.9×ASM+97.2????? (17)

    where B5 represents the biomass of wheat in the pre-winter period? combining VARI? and ASM. The R2was 0.565, and the accuracy was slightly improvedcompared with the single color index model (2.17%increase).

    (2) Estimation model of wheat biomass in thejointing period

    From the pre-wintering period to the jointingperiod, the UAV image texture indices did not im‐prove significantly. Among the four texture featureindices, only two had a significant correlation withbiomass, with COR exhibiting the highest correla‐tion with an r of 0.574(Table 4). Therefore, the tex‐ture feature index COR was combined with the col‐or index ExGR. The wheat biomass estimation mul‐tiple regression model is:

    B6=8762.5×ExGR?259.4×COR+2075.8? (18)

    where? B6? represents? the? biomass? of wheat? in? thejointing period combining ExGR and COR. The R2was 0.833,? and? the? accuracy? was? improved? com ‐pared with the single color index model (3.61% in ‐crease).

    (3) Estimation model of wheat biomass in thebooting period

    The? colors? of the? wheat? UAV? images? in? thebooting period showed a certain saturation phenom ‐enon, and the texture feature indices were also af‐fected. Among the four texture feature indices, onlyCON? was? significantly? correlated? with? biomass,with an r of 0.417(Table 4). Therefore, the texturefeature index CON was combined with the color in ‐dex MGRVI, and the wheat biomass estimation mul‐tiple regression model is:

    B7=24027.4×MGRVI?3098.6×CON+2252.3(19)

    where? B7? represents? the? biomass? of wheat? in? the booting period? combining MGRVI? and ?CON. The R2 was 0.762, and the accuracy was significantly im‐ proved compared with the single color index model (13.73% increase).

    (4) Estimation model of wheat biomass in the flowering period

    The? correlation? between? UAV? image? texture feature indices and wheat biomass in the flowering period? was? relatively? low,? and? the? correlation? be‐ tween the four feature indices and wheat biomass did not reach a significant level (Table 4). Therefore, the index ASM with the highest correlation coefficient was? selected? and? combined? with? the? color? index VARI. The wheat biomass? estimation? multiple re‐gression model is:

    B8=42654.9×VARI?5595.3×ASM+8507.9(20) where? B8? represents? the? biomass? of wheat? in? the flowering period? combining VARI? and ASM. The R2 was 0.485. The accuracy was much higher than the single color index model (5.21% increase).

    3.3.2 Model validation

    (1) Validation of the wheat biomass estimation model in the pre-wintering period

    The independently measured data were used to validate the wheat biomass estimation model in the pre-wintering period,? and the 1:1 relationship be‐ tween the measured values? and the model predic‐tions? was? plotted (Figure 2(a)). There? was? good agreement? between? the? predicted? value? of wheat biomass and the measured value in the pre-winter‐ing? period. The? predicted? R2? of? the? model? was 0.571, which was better than the single color index model? (6.13%? increase).? Correlation? analysis showed? that the? correlation between? the predicted value and the measured value was significant, indi‐cating that the estimation model is feasible. In addi‐tion, the simulated RMSE was 25.49 kg/ha, whichwas smaller than the RMSE of the single color in ‐dex model (reduced by 8.57%), indicating that thereliability of the model had been further improved.

    (2) Validation of the wheat biomass estimationmodel in the jointing period

    The? validation? results? of the jointing? estima‐tion model are? shown in Fig.2(b). It can be? seenthat? the? predicted? value? of wheat? biomass? in? thejointing? period? was? close? to? the? measured? value,and there was a good agreement between the two.The predicted R2 of the model was 0.658, which isbetter than the pre-wintering period, and the predic‐tion accuracy was improved compared to the singlecolor? index? model (4.28% increase). Correlationanalysis? indicated that there was? a? correlation be‐tween the predicted value and the measured value,indicating that the estimation model is feasible. Thesimulated RMSE was 443.20 kg/ha, which was sig ‐nificantly smaller than the RMSE of the single col‐or index model (14.27% decrease), indicating thatthe? reliability? of the? model? had? been? greatly? im‐proved.

    (3) Validation of the wheat biomass estimationmodel in the booting period

    The validation results of the booting period es‐timation model are shown in Fig.2(c). The predict‐ed? value? of wheat? biomass? in? the? booting? periodwas close to the measured value, and there was highconsistency between the two. The model predictedan R2 of 0.753, which was better than the jointingperiod, and the prediction accuracy was greatly im‐proved? compared? to? the? single? color? index? model(6.36% increase). Correlation? analysis? indicatedthat the correlation between the predicted value andthe measured value was significant, indicating thatthe? estimation? model? is? feasible. In? addition,? thesimulated? RMSE? was 816.25 kg/ha,? which? wassmaller? than? the? RMSE? of the? single? color? indexmodel (5.99% decrease), indicating that the reliabili‐ ty of the model had been improved.

    (4) Validation of the wheat biomass estimation model in the flowering period

    The validation results of the flowering period estimation model are shown in Fig.2(d). The pre‐dicted value of wheat biomass in the booting period was close to the measured value, and there was a high agreement between the two. The predicted R2 of the model was 0.515, and the prediction accuracy was? improved? compared to the? single? color? index model (10.99% increase). Correlation analysis indi‐cated that the correlation between the predicted val‐ue and the measured value was significant, indicat‐ing that the? estimation model? is? feasible. In? addi‐tion,? the? simulated? RMSE? was 1396.97 kg/ha, which was smaller than the RMSE of the single col‐ or index model (reduced by 9.28%), indicating that the reliability of the model had been improved.

    Based on the above results, the biomass estima‐tion? model? of wheat? in? different? growth? periods based on the combination of UAV image color and texture? feature? indices? was? significantly? improved over the single color index model and can be used to estimate the biomass of wheat in different growth periods.

    4? Discussion

    At present, few studies have monitored growth indices such as wheat biomass using the RGB im‐ age data of UAVs. Hunt et al.[26] obtained wheat can ‐opy images using UAVs and used the vegetation in ‐ dices NDVI and GNDVI to monitor wheat growth in order to validate the availability of UAV images in wheat growth monitoring. In this study, the RGB images collected in the four key growth periods of wheat were systematically analyzed. The image in ‐ formation was? extracted,? and? eight? common? colorindices and four texture feature indices were calcu‐lated. The best performing color index and texture feature index were selected using correlation analy‐ sis. The results showed that the biomass estimation model constructed using a single color index could reduce the estimation accuracy due to the saturation of the image in the late growth period of wheat. In order to solve the above problems, the UAV image color and texture feature indices were combined to establish a multivariate model of wheat biomass es‐timation? based? on? two? types? of? indices,? and? the model? was? tested? using? independently ?measured biomass? data,? achieving? an? overall? better? perfor‐mance than a single color index model.

    Other studies have combined image color indi‐ces with other indices to estimate crop yield or bio ‐ mass. For example, Lu et al.[27] used UAV image da‐ ta and point cloud data to estimate the aboveground biomass? of wheat. Compared? to? the use? of single color? data,? the? model? with? combined? indices? im‐ proved the estimation accuracy, R2 was 0.78, RMSE was 1.34 t/ha and rRMSE was 28.98%. Duan et al.[28] established a new method combining the vegetation indices based on UAV images and the abundance in ‐ formation? obtained? from? spectral? mixture? analysis (SMA). The results showed that the vegetation in ‐dex? with? abundance? information? exhibited ?better predictive ability for rice yield than the vegetation index? alone? with? the? coefficient? of determination reaching 0.6 and estimation error below 10%. Zhou et al.[29] established wheat yield prediction model by using color and texture feature indices of RGB im‐ ages at wheat booting and flowering stages, and the R2 of the model increased by 2.15% and 3.69% com ‐ pared with the single-color index model, respective‐ly. The? above? results? demonstrate? that? combining the color indices of UAV images with different indi‐ces,? such? as? shape, texture,? and? coverage? can? im‐prove the estimation effect of the model.

    In addition, previous studies often only moni‐tored the biomass in a certain growth period of thecrop due to image availability or other constraints,and thus the results? are difficult to? extrapolate. Inthis study, UAV images of the four key growth peri‐ods of wheat were analyzed, and the dynamic moni‐toring model of wheat biomass based on the UAVplatform was established. The model was tested us‐ing? independently? measured? data,? and? the? resultswere all reliable, indicating that it is feasible to dy‐namically monitor wheat biomass using a UAV plat‐form equipped with a digital camera, which is suit‐able for small and medium area applications. How ‐ever,? in? the? future,? UAVs? will? be? equipped? withmore? types? of? cameras. Therefore,? in? future? re‐search, information such as the spectrum, color, andtexture of the image could be comprehensively usedto improve the monitoring accuracy and range.

    5 Conclusions

    In this study, estimation models of wheat bio ‐mass at main growth stages were constructed basedon the color and texture index of UAV images.

    Firstly, the color index and texture feature in ‐dex optimally correlated with wheat biomass wereidentified. Among the eight color indices, most ofthe? correlations? with? wheat? biomass? were? signifi‐cant, and the correlation of VARI in the pre-winter‐ing period was the highest at r=0.743**. The corre‐lation of ExGR in the jointing period was the high ‐est at r=0.911**. The correlation of MGRVI in thebooting period was hight at r=0.817**, and the cor‐relation of VARI in the flowering period was hightat r=0.679**. The correlation between the four im‐age texture feature indices and wheat biomass waspoor, and only a few indices reached significant cor‐relation levels.

    Then, using the color indices of VARI, ExGR, MGRVI and VARI, the biomass estimation models of wheat in the pre-wintering, jointing, booting, and flowering periods were constructed, and the wheat biomass? estimation multivariate models were? con ‐ structed? by? combining? the? texture? feature? indices that? with? the? highest? correlation? with? wheat? bio ‐ mass.

    Finally, the models were validated using inde‐pendently? measured? biomass? data. The? results showed? that both? two? types? of estimation? models passed? the? significance? test,? with? the? correlation reaching? a? significant? level? and? the? RMSE? being small. The? wheat biomass? estimation? model? com ‐bined with the UAV image color and texture feature indices was better than the single color index model.

    References:

    [1] WANG? E. Study? on? remote? sensing? monitoring? ofmaize growth and biomass[D]. Hefei: Anhui Agricul‐tural University, 2018.

    [2] FAN Y, GONG Z, ZHAO W, et al. Study on vegetationbiomass? inversion method based? on hyperspectral re‐ mote? sensing[J]. Journal? of Hebei Normal University (Natural Science Edition), 2016, 3:267-271.

    [3] JIMENEZ-SIERRA? D,? CORREA? E,? BEN?TEZ-RE ‐STREPO? H,? et? al. Novel? feature-extraction? methods for? the? estimation? of? above-ground? biomass? in? rice crops[J]. Sensors, 2021, 21(13):1-14.

    [4] HOU X, NIU Z, HUANG N, et al. The hyperspectralremote sensing estimation models of total biomass and true LAI of wheat[J]. Remote Sensing for Land & Re‐ sources, 2012, 24(4):30-35.

    [5] LIU Q, PENG D, TU Y, et al. Estimating forest bio‐mass by partial least squares regression[J]. Journal of Northeast Forestry University, 2014, 7:44-47.

    [6] YUE J, ZHOU C, GUO W, et al. Estimation of winter-wheat above-ground biomass using the wavelet analy‐ sis? of unmanned? aerial? vehicle-based? digital? images and hyperspectral crop canopy images[J]. International Journal of Remote Sensing, 2021, 42(5):1602-1622.

    [7] LI? J,? SHI Y, VEERANAMPALAYAM-SIVAKUMARA N, et al. Elucidating sorghum biomass, nitrogen and chlorophyll? contents with? spectral? and? morphological traits derived from unmanned aircraft system[J]. Fron‐tiers in Plant Science, 2018, 9:1-12.

    [8] GU Z, JU W, LI L, et al. Using vegetation indices andtexture measures to estimate vegetation fractional cov‐erage (VFC) of planted and natural forests in Nanjingcity, China[J]. Advances in Space Research, 2013, 51(7):1186-1194.

    [9] SARKER L R, NICHOL J E. Improved forest biomassestimates using ALOS AVNIR-2 texture indices[J]. Re‐mote Sensing of Environment, 2011, 115(4):968-977.

    [10] CAO Q, XU D, JU H. The biomass estimation of man‐grove? community? based? on? the? textural? features? andspectral information of TM images[J]. Forest Resourc‐es Management, 2010, 6:102-108.

    [11] MU? Q,? GAO? Z, BAO Y,? et? al. Estimation? of sparsevegetation biomass based on grey-level co-occurrencematrix of vegetation indices[J]. Remote Sensing Infor‐mation, 2016, 1:58-63.

    [12] LIU C, YANG G, LI Z, et al. Biomass estimation inwinter wheat by UAV spectral information and textureinformation? fusion[J]. Scientia? Agricultura? Sinica,2018, 51(16):3060-3073.

    [13] LEE K J, LEE B W. Estimation of rice growth and ni‐trogen nutrition status using color digital camera imageanalysis[J]. European Journal of Agronomy, 2013, 48(3), 57-65.

    [14] WANG Y, WANG D, ZHANG G, et al. Estimating ni‐trogen status of rice using the image segmentation ofG-R? thresholding? method[J]. Field? Crops? Research,2013, 149(149):33-39.

    [15] WANG J, ZHOU Q, SHANG J, et al. UAV- and ma‐chine learning-based retrieval of wheat SPAD values atthe? overwintering? stage? for? variety? screening[J]. Re‐mote sensing, 2021, 13:1-19.

    [16] YUE J, FENG H, LI Z, et al. Mapping winter-wheatbiomass? and? grain? yield based? on? a? crop? model? andUAV? remote? sensing[J]. International? Journal? of Re‐mote Sensing, 2021, 42(5):1577-1601.

    [17] SHAN C, LIAO S, GONG Y, et al. Application of digi‐tal image processing for determination of vertical dis‐tribution of biomass in the canopy of winter wheat[J].Acta AgronomicaSinica, 2007, 33(3):419-424.

    [18] CHEN W. Evsluation of seed emergence unifermity ofwheat based on UAV image[D]. Yangzhou: YangzhouUnivefrsity, 2018.

    [19] GITELSON A A, STARK R, GRITS U, et al. Vegeta‐tion and soil lines in visible spectral space: A conceptand technique for remote estimation of vegetation frac‐tion[J]. International Journal of Remote Sensing, 2002,23(13):2537-2562.

    [20] MAO D, WU X, DEPPONG C, et al. Negligible roleof antibodies and C5 in pregnancy loss associated ex‐clusively with C3-dependent mechanisms through com ‐plement? alternative? pathway[J]. Immunity, 2003, 19(6):813-822.

    [21] WOEBBECKE D M, MEYER G E, VON BARGENK, et al. Plant species identification, size, and enumera‐tion? using? machine? vision? techniques? on? near-binary images[J]. Proceedings of SPIE-The International Soci‐ety for Optical Engineering, 1993, 1836:1-12.

    [22] HUNT JR E R, DAUGHTRY C S T, EITEL J U H, etal. Remote sensing leaf chlorophyll content using a vis‐ible? band? index[J]. Agronomy? Journal, 2011, 103(4):1090-1099.

    [23] TUCKER P W, HAZEN E E, COTTON F A. Staphylo‐coccal nuclease reviewed: A prototypic? study in? con‐ temporary? enzymology[J]. Molecular & Cellular Bio‐ chemistry, 1979, 23(2):67-86.

    [24] BENDIG J, YU K, AASEN H, et al. Combining UAV-based plant height from crop? surface models, visible, and near infrared vegetation indices for biomass moni‐toring? in? barley[J]. International? Journal? of Applied Earth Observation & Geoinformation, 2015, 39:79-87.

    [25] HARALICK R M,? SHANMUGAM K, DINSTEIN I.Textural? features? for? image? classification[J]. IEEETransactions on Systems Man & Cybernetics, 1973, 3(6):610-621.

    [26] HUNT JR E R , HIVELY W D, FUJIKAWA S, et al.Acquisition? of? NIR-Green-Blue? digital? photographsfrom? unmanned? aircraft? for? crop? monitoring[J]. Re‐mote Sensing, 2010, 2(1):290-305.

    [27] LU N, ZHOU J, HAN Z,? et? al. Improved? estimationofaboveground biomass? in wheat? from RGB? imageryand point cloud data acquired withalow?cost unmannedaerial vehicle system[J]. Plant Methods, 2019, 15:1-16.

    [28] DUAN B, FANG S, ZHU R, et al. Remote estimationof rice yield with unmanned aerial vehicle (UAV) dataand spectral mixture analysis[J]. Frontiers in Plant Sci‐ence, 2019, 10:1-14.

    [29] ZHOU? Y,? WANG? D,? CHEN? C,? et? al. Prediction? ofwheat yield based on color index and texture feature in‐dex of UAV RGB image[J]. Journal of Yangzhou Uni‐versity (Agricultural and Life? Science Edition), 2021,42(3):110-116.

    猜你喜歡
    生物量小麥
    主產區(qū)小麥收購進度過七成
    小麥測產迎豐收
    小麥春季化控要掌握關鍵技術
    孔令讓的“小麥育種夢”
    金橋(2021年10期)2021-11-05 07:23:28
    葉面施肥實現(xiàn)小麥畝增產83.8千克
    輪牧能有效促進高寒草地生物量和穩(wěn)定性
    哭娃小麥
    不同NPK組合對芳樟油料林生物量的影響及聚類分析
    草地生物量遙感估算方法綜述
    甘肅科技(2020年21期)2020-04-13 00:33:46
    生物量高的富鋅酵母的開發(fā)應用
    亚洲精品在线美女| 丰满少妇做爰视频| 亚洲国产欧美在线一区| 亚洲国产av影院在线观看| 麻豆国产av国片精品| 可以免费在线观看a视频的电影网站| 99精品在免费线老司机午夜| 一区二区三区激情视频| 欧美变态另类bdsm刘玥| cao死你这个sao货| 久久精品国产99精品国产亚洲性色 | 亚洲国产成人一精品久久久| 男男h啪啪无遮挡| 久久免费观看电影| av在线播放免费不卡| 丝瓜视频免费看黄片| 丝袜在线中文字幕| 三上悠亚av全集在线观看| 久久av网站| 色精品久久人妻99蜜桃| 熟女少妇亚洲综合色aaa.| 久久精品亚洲精品国产色婷小说| 丝瓜视频免费看黄片| 美女高潮到喷水免费观看| 欧美乱妇无乱码| 国产aⅴ精品一区二区三区波| 亚洲午夜精品一区,二区,三区| 国产成人系列免费观看| 欧美激情高清一区二区三区| 极品人妻少妇av视频| 真人做人爱边吃奶动态| 成人av一区二区三区在线看| 91精品三级在线观看| 国产成人影院久久av| 法律面前人人平等表现在哪些方面| 黑人操中国人逼视频| 久久久久久久国产电影| 欧美黑人欧美精品刺激| 大香蕉久久成人网| www.精华液| 国产亚洲精品一区二区www | 19禁男女啪啪无遮挡网站| 青草久久国产| 9色porny在线观看| 九色亚洲精品在线播放| 久久精品国产亚洲av香蕉五月 | 欧美中文综合在线视频| 成年人午夜在线观看视频| 欧美精品高潮呻吟av久久| 黄色视频,在线免费观看| 人人妻人人澡人人看| 亚洲欧美日韩高清在线视频 | 国产精品一区二区免费欧美| 黄色视频在线播放观看不卡| 精品一品国产午夜福利视频| 99国产综合亚洲精品| 久久人妻熟女aⅴ| 国产精品影院久久| av免费在线观看网站| 亚洲成人手机| 国产国语露脸激情在线看| a在线观看视频网站| 国产av又大| 夜夜夜夜夜久久久久| 欧美精品一区二区免费开放| 亚洲午夜理论影院| 久久久久久亚洲精品国产蜜桃av| 亚洲国产av新网站| e午夜精品久久久久久久| 777米奇影视久久| 成人国产av品久久久| 久久久久久人人人人人| 性少妇av在线| 一级片免费观看大全| 国产亚洲精品久久久久5区| 18禁美女被吸乳视频| 成在线人永久免费视频| av超薄肉色丝袜交足视频| 一边摸一边抽搐一进一小说 | 久久国产精品人妻蜜桃| 久久久久久免费高清国产稀缺| 亚洲第一青青草原| 中文欧美无线码| 91老司机精品| 在线观看免费视频日本深夜| 亚洲男人天堂网一区| 99riav亚洲国产免费| 男女高潮啪啪啪动态图| 麻豆成人av在线观看| 在线永久观看黄色视频| 欧美在线一区亚洲| 午夜激情久久久久久久| 男女边摸边吃奶| 欧美成狂野欧美在线观看| 欧美精品av麻豆av| 亚洲色图av天堂| av天堂在线播放| 国产日韩欧美视频二区| 亚洲精品一卡2卡三卡4卡5卡| 久久久久精品人妻al黑| 高潮久久久久久久久久久不卡| 男女之事视频高清在线观看| 国产91精品成人一区二区三区 | 日韩大码丰满熟妇| 久久中文字幕一级| 日韩视频一区二区在线观看| 美女高潮到喷水免费观看| 十八禁网站免费在线| 午夜福利在线免费观看网站| 国产熟女午夜一区二区三区| 一级毛片女人18水好多| 天天躁夜夜躁狠狠躁躁| 免费高清在线观看日韩| 亚洲精品乱久久久久久| 欧美人与性动交α欧美软件| 巨乳人妻的诱惑在线观看| 男女免费视频国产| √禁漫天堂资源中文www| 亚洲自偷自拍图片 自拍| 久久精品国产a三级三级三级| 久久国产精品大桥未久av| 女警被强在线播放| 精品久久久精品久久久| 久久国产精品男人的天堂亚洲| 国产成人精品久久二区二区91| 国产精品免费大片| 国产视频一区二区在线看| 精品乱码久久久久久99久播| 美女主播在线视频| 午夜91福利影院| 日韩成人在线观看一区二区三区| 别揉我奶头~嗯~啊~动态视频| av又黄又爽大尺度在线免费看| 午夜91福利影院| 波多野结衣av一区二区av| 肉色欧美久久久久久久蜜桃| 午夜久久久在线观看| 又紧又爽又黄一区二区| 人人妻人人爽人人添夜夜欢视频| 日韩有码中文字幕| 精品久久久精品久久久| 欧美黑人精品巨大| 国产色视频综合| 日日摸夜夜添夜夜添小说| 日韩欧美国产一区二区入口| 亚洲欧美一区二区三区黑人| 高清在线国产一区| 国产成人欧美| 久久国产精品大桥未久av| 精品午夜福利视频在线观看一区 | 午夜福利在线免费观看网站| 欧美日韩亚洲国产一区二区在线观看 | 一区在线观看完整版| 50天的宝宝边吃奶边哭怎么回事| 老熟女久久久| 18禁美女被吸乳视频| 午夜久久久在线观看| 动漫黄色视频在线观看| 老司机影院毛片| 成年女人毛片免费观看观看9 | 悠悠久久av| 国产免费现黄频在线看| 777米奇影视久久| 一级a爱视频在线免费观看| 亚洲av美国av| 国产一区二区在线观看av| 热99国产精品久久久久久7| 日韩欧美一区视频在线观看| 男人操女人黄网站| 丰满少妇做爰视频| 午夜福利一区二区在线看| 成年人黄色毛片网站| av国产精品久久久久影院| 亚洲av成人不卡在线观看播放网| 亚洲国产精品一区二区三区在线| 久久久久久久久免费视频了| 人人妻,人人澡人人爽秒播| 狠狠狠狠99中文字幕| 亚洲精品国产色婷婷电影| 国产无遮挡羞羞视频在线观看| 久久久国产精品麻豆| 777米奇影视久久| 国产亚洲一区二区精品| 亚洲av日韩精品久久久久久密| 成人国语在线视频| 9热在线视频观看99| 国产不卡av网站在线观看| 国产欧美日韩一区二区精品| 午夜91福利影院| 80岁老熟妇乱子伦牲交| 在线观看免费午夜福利视频| av不卡在线播放| 欧美变态另类bdsm刘玥| 国产黄频视频在线观看| 国产成人精品久久二区二区免费| 国产日韩欧美在线精品| 久久精品成人免费网站| 热re99久久国产66热| 欧美在线黄色| 波多野结衣一区麻豆| 精品高清国产在线一区| av视频免费观看在线观看| 成人手机av| 王馨瑶露胸无遮挡在线观看| 少妇被粗大的猛进出69影院| 亚洲人成电影免费在线| 欧美激情久久久久久爽电影 | 少妇猛男粗大的猛烈进出视频| 新久久久久国产一级毛片| 国产精品久久久久久人妻精品电影 | 丝袜喷水一区| 日韩大码丰满熟妇| 咕卡用的链子| 国产精品免费一区二区三区在线 | 婷婷丁香在线五月| 精品国产亚洲在线| 极品教师在线免费播放| 成人18禁高潮啪啪吃奶动态图| 亚洲专区中文字幕在线| 男女下面插进去视频免费观看| 两个人看的免费小视频| 丝袜喷水一区| 亚洲色图 男人天堂 中文字幕| 久久精品国产综合久久久| 在线观看免费视频网站a站| 麻豆国产av国片精品| 免费在线观看视频国产中文字幕亚洲| 日本精品一区二区三区蜜桃| 成人国语在线视频| 热re99久久精品国产66热6| 极品教师在线免费播放| 黑人欧美特级aaaaaa片| 91成人精品电影| 中国美女看黄片| 午夜免费鲁丝| av在线播放免费不卡| 成年人黄色毛片网站| 搡老熟女国产l中国老女人| 国产淫语在线视频| 精品国产乱码久久久久久小说| 欧美激情久久久久久爽电影 | 亚洲午夜理论影院| 两个人免费观看高清视频| 亚洲熟女精品中文字幕| 一本大道久久a久久精品| 国产欧美日韩一区二区三| 99久久国产精品久久久| 日韩中文字幕欧美一区二区| 叶爱在线成人免费视频播放| 久久午夜综合久久蜜桃| 欧美+亚洲+日韩+国产| 久久人妻熟女aⅴ| 两个人看的免费小视频| 亚洲欧美色中文字幕在线| 国产不卡av网站在线观看| 国产欧美日韩一区二区精品| 岛国在线观看网站| 国产在线免费精品| 国产人伦9x9x在线观看| 亚洲精品在线观看二区| 超色免费av| 黑人操中国人逼视频| 亚洲精品久久成人aⅴ小说| 电影成人av| 欧美黄色淫秽网站| 精品午夜福利视频在线观看一区 | 纵有疾风起免费观看全集完整版| 又大又爽又粗| 久久天躁狠狠躁夜夜2o2o| 操出白浆在线播放| 国产精品 欧美亚洲| 成人永久免费在线观看视频 | 久久香蕉激情| 搡老乐熟女国产| 夜夜骑夜夜射夜夜干| 狠狠精品人妻久久久久久综合| 精品乱码久久久久久99久播| 久久免费观看电影| 一边摸一边抽搐一进一出视频| h视频一区二区三区| 久久久精品国产亚洲av高清涩受| 妹子高潮喷水视频| 久久免费观看电影| 国产老妇伦熟女老妇高清| 亚洲精品美女久久久久99蜜臀| av网站在线播放免费| 三上悠亚av全集在线观看| 精品国产一区二区三区久久久樱花| 最新美女视频免费是黄的| 亚洲欧美色中文字幕在线| 国产成人av教育| 9热在线视频观看99| 女性生殖器流出的白浆| 精品少妇久久久久久888优播| avwww免费| e午夜精品久久久久久久| 丰满饥渴人妻一区二区三| 国产亚洲精品久久久久5区| 国产1区2区3区精品| 国产精品国产av在线观看| 日本wwww免费看| 在线观看www视频免费| 亚洲精品中文字幕一二三四区 | 51午夜福利影视在线观看| 午夜免费鲁丝| a级片在线免费高清观看视频| 操美女的视频在线观看| 老司机靠b影院| 母亲3免费完整高清在线观看| 精品高清国产在线一区| 欧美黑人欧美精品刺激| 色尼玛亚洲综合影院| 天天躁夜夜躁狠狠躁躁| 免费人妻精品一区二区三区视频| 一边摸一边抽搐一进一出视频| 老司机影院毛片| 国产成人av激情在线播放| 天天影视国产精品| 亚洲人成伊人成综合网2020| 亚洲国产成人一精品久久久| 中国美女看黄片| 在线十欧美十亚洲十日本专区| kizo精华| 欧美精品av麻豆av| 国产一区有黄有色的免费视频| 男人操女人黄网站| 欧美成人午夜精品| 国产精品 国内视频| 美国免费a级毛片| 在线天堂中文资源库| 欧美中文综合在线视频| 丁香六月天网| 老司机午夜十八禁免费视频| 久久久久久人人人人人| 亚洲熟女精品中文字幕| 久久久久久免费高清国产稀缺| 叶爱在线成人免费视频播放| 美女福利国产在线| 欧美黑人欧美精品刺激| 国产精品免费大片| 国产不卡一卡二| av又黄又爽大尺度在线免费看| www.自偷自拍.com| 啪啪无遮挡十八禁网站| 国产aⅴ精品一区二区三区波| 美女视频免费永久观看网站| 国产精品免费一区二区三区在线 | 欧美精品高潮呻吟av久久| 亚洲成av片中文字幕在线观看| 日韩有码中文字幕| 9191精品国产免费久久| videos熟女内射| 电影成人av| 国产精品一区二区免费欧美| 伊人久久大香线蕉亚洲五| 交换朋友夫妻互换小说| 淫妇啪啪啪对白视频| 国产亚洲精品第一综合不卡| 99riav亚洲国产免费| 亚洲午夜理论影院| 久久久精品免费免费高清| 精品国产乱码久久久久久小说| 国产精品98久久久久久宅男小说| 亚洲精品国产精品久久久不卡| 成年女人毛片免费观看观看9 | 国产日韩欧美在线精品| 欧美一级毛片孕妇| 国产不卡一卡二| 一级片'在线观看视频| 久久久国产欧美日韩av| 狂野欧美激情性xxxx| 丁香六月天网| 狂野欧美激情性xxxx| 欧美av亚洲av综合av国产av| 日韩中文字幕视频在线看片| avwww免费| 大片电影免费在线观看免费| 欧美在线黄色| 黑人猛操日本美女一级片| 日韩视频在线欧美| 777米奇影视久久| 国产真人三级小视频在线观看| 精品欧美一区二区三区在线| h视频一区二区三区| 亚洲avbb在线观看| 国产av一区二区精品久久| 亚洲国产中文字幕在线视频| 午夜福利乱码中文字幕| www.熟女人妻精品国产| 欧美老熟妇乱子伦牲交| 人人妻人人添人人爽欧美一区卜| 亚洲久久久国产精品| 亚洲 欧美一区二区三区| 91精品三级在线观看| 亚洲欧美日韩高清在线视频 | 王馨瑶露胸无遮挡在线观看| 俄罗斯特黄特色一大片| 国产亚洲精品一区二区www | 精品久久蜜臀av无| 亚洲第一av免费看| 美女扒开内裤让男人捅视频| 日韩精品免费视频一区二区三区| 免费久久久久久久精品成人欧美视频| 欧美变态另类bdsm刘玥| 久久热在线av| 十八禁网站免费在线| 国产在线精品亚洲第一网站| 黄色视频不卡| 99久久99久久久精品蜜桃| 国产午夜精品久久久久久| xxxhd国产人妻xxx| 久久精品人人爽人人爽视色| 美国免费a级毛片| 极品教师在线免费播放| 日韩免费高清中文字幕av| 极品教师在线免费播放| 国产成人精品在线电影| 色尼玛亚洲综合影院| 另类亚洲欧美激情| 精品一区二区三区av网在线观看 | 欧美另类亚洲清纯唯美| 岛国毛片在线播放| 亚洲国产欧美网| 精品国产一区二区三区久久久樱花| 我要看黄色一级片免费的| 极品教师在线免费播放| 免费日韩欧美在线观看| 久久精品成人免费网站| 国产亚洲一区二区精品| 男人操女人黄网站| av线在线观看网站| 久久精品国产99精品国产亚洲性色 | 露出奶头的视频| 窝窝影院91人妻| 久久精品人人爽人人爽视色| 丰满少妇做爰视频| 精品免费久久久久久久清纯 | 国产av一区二区精品久久| 亚洲欧洲精品一区二区精品久久久| 国产不卡av网站在线观看| 久久人人爽av亚洲精品天堂| 国产亚洲欧美在线一区二区| 久久99热这里只频精品6学生| 国产免费av片在线观看野外av| 久久久精品94久久精品| 精品乱码久久久久久99久播| 一区二区三区国产精品乱码| 中文字幕人妻丝袜制服| a级毛片黄视频| 久久天躁狠狠躁夜夜2o2o| 欧美日韩黄片免| 精品国产乱码久久久久久小说| www.自偷自拍.com| 午夜激情久久久久久久| 国产精品免费一区二区三区在线 | 高清欧美精品videossex| 少妇裸体淫交视频免费看高清 | 成人影院久久| 高清视频免费观看一区二区| 精品少妇内射三级| 一夜夜www| 国产精品1区2区在线观看. | 亚洲七黄色美女视频| 日本a在线网址| √禁漫天堂资源中文www| 丁香欧美五月| 欧美黄色淫秽网站| 日韩制服丝袜自拍偷拍| 欧美日韩成人在线一区二区| 久久久久网色| 黑丝袜美女国产一区| 窝窝影院91人妻| 欧美激情 高清一区二区三区| 午夜久久久在线观看| 日韩成人在线观看一区二区三区| 欧美人与性动交α欧美精品济南到| 变态另类成人亚洲欧美熟女 | 午夜福利一区二区在线看| 国产av又大| 国产精品一区二区精品视频观看| 香蕉丝袜av| 亚洲av片天天在线观看| 老汉色av国产亚洲站长工具| 国产片内射在线| avwww免费| 日韩中文字幕视频在线看片| 美女午夜性视频免费| 久久久精品94久久精品| 19禁男女啪啪无遮挡网站| 一本久久精品| 汤姆久久久久久久影院中文字幕| 欧美 日韩 精品 国产| 两人在一起打扑克的视频| 日韩精品免费视频一区二区三区| 久久青草综合色| 丝袜美腿诱惑在线| 麻豆成人av在线观看| 99精品久久久久人妻精品| 色在线成人网| 搡老乐熟女国产| 亚洲中文av在线| 亚洲专区字幕在线| 午夜福利在线观看吧| 女人被躁到高潮嗷嗷叫费观| 成人国语在线视频| 欧美黄色片欧美黄色片| 午夜日韩欧美国产| 纵有疾风起免费观看全集完整版| 蜜桃国产av成人99| 可以免费在线观看a视频的电影网站| 国产免费av片在线观看野外av| 亚洲 欧美一区二区三区| 男女免费视频国产| 一区二区三区激情视频| 午夜老司机福利片| 日韩视频在线欧美| 50天的宝宝边吃奶边哭怎么回事| 午夜福利在线免费观看网站| 欧美精品亚洲一区二区| 波多野结衣一区麻豆| 久久久国产欧美日韩av| 黄色丝袜av网址大全| 一二三四社区在线视频社区8| 免费一级毛片在线播放高清视频 | av免费在线观看网站| 大香蕉久久成人网| 18禁观看日本| 亚洲综合色网址| 99国产精品免费福利视频| 9191精品国产免费久久| 久久亚洲真实| 超碰97精品在线观看| 欧美黄色淫秽网站| 91麻豆精品激情在线观看国产 | 久久国产精品人妻蜜桃| 国产三级黄色录像| 9热在线视频观看99| 国产极品粉嫩免费观看在线| 欧美精品高潮呻吟av久久| 国产高清国产精品国产三级| 精品久久久久久久毛片微露脸| 色老头精品视频在线观看| 国产精品成人在线| 97在线人人人人妻| 搡老乐熟女国产| 免费高清在线观看日韩| 水蜜桃什么品种好| 亚洲一卡2卡3卡4卡5卡精品中文| 日本欧美视频一区| 丁香六月欧美| 国产成人精品久久二区二区91| 国产成人欧美| 国产成人啪精品午夜网站| 久久久久精品人妻al黑| 下体分泌物呈黄色| 欧美人与性动交α欧美精品济南到| 国产又色又爽无遮挡免费看| 国精品久久久久久国模美| cao死你这个sao货| 久久久久久久国产电影| 国产精品免费大片| 女性生殖器流出的白浆| 国产视频一区二区在线看| 性少妇av在线| 亚洲第一青青草原| 咕卡用的链子| 亚洲熟女精品中文字幕| 精品国产一区二区三区四区第35| 99精品在免费线老司机午夜| 男女免费视频国产| 纯流量卡能插随身wifi吗| 免费av中文字幕在线| 极品教师在线免费播放| 久久久精品区二区三区| av网站在线播放免费| 捣出白浆h1v1| 9热在线视频观看99| 极品人妻少妇av视频| 91成人精品电影| 国产精品久久电影中文字幕 | 亚洲一码二码三码区别大吗| 国产成人精品久久二区二区免费| 国产免费现黄频在线看| 久久精品亚洲精品国产色婷小说| 少妇裸体淫交视频免费看高清 | av不卡在线播放| 麻豆av在线久日| 国产高清视频在线播放一区| aaaaa片日本免费| 热99久久久久精品小说推荐| 日本av手机在线免费观看| 十八禁网站免费在线| 午夜福利视频精品| 国产精品自产拍在线观看55亚洲 | 午夜激情av网站| 两性夫妻黄色片| 91字幕亚洲| 在线观看免费午夜福利视频| 亚洲国产毛片av蜜桃av| 天天影视国产精品| 91大片在线观看| 亚洲精品粉嫩美女一区| 搡老熟女国产l中国老女人| 夜夜爽天天搞| 在线观看www视频免费| 男人操女人黄网站| 日韩欧美国产一区二区入口| 老司机福利观看| 国产精品国产av在线观看| 国产成人av激情在线播放| 日韩视频一区二区在线观看| 不卡av一区二区三区| 一区二区三区精品91| 色在线成人网| 国产高清videossex|