謝俊+郭春裕+張鵬+李威霖
摘 要:為了檢測電氣產(chǎn)品的安全狀態(tài),采用紫外成像檢測技術(shù)對電氣產(chǎn)品的放電狀態(tài)進行研究。紫外圖像在采集的時候難免會受到各種各樣的干擾和噪聲,大的紫外光斑周圍有很多微小的白色光斑,這些光斑會對紫外圖像特征量的提取產(chǎn)生嚴重的影響,需要通過預處理來濾除這些干擾。一般來說,圖像噪聲的來源有以下三方面:一為光電、電磁轉(zhuǎn)換過程中引入的噪聲;二為電氣產(chǎn)品本身存在的強電磁脈沖的干擾;三為自然起伏性噪聲。這些噪聲導致紫外圖像不能符合后續(xù)的存儲和處理要求。此時就需要對其進行預處理來消除干擾和噪聲的影響,從而抑制與實際信號無關(guān)的雜波,提高對后續(xù)圖像的處理能力和精確度。
關(guān)鍵詞:圖像預處理;方法;精確度
中圖分類號:TP391.4 文獻標志碼:A 文章編號:2095-2945(2018)07-0018-03
Abstract: In order to detect the safe state of electrical products, the discharge state of electrical products is studied by ultraviolet (UV) imaging technology. UV images will inevitably be subjected to a variety of interference and noise, large ultraviolet spots around a lot of small white spots, which have a serious impact on the extraction of ultraviolet image features. These disturbances need to be filtered out by pre-processing. Generally speaking, the source of image noise has the following three aspects: the first is the noise introduced in the process of photoelectric and electromagnetic conversion; the second is the interference of the strong electromagnetic pulse existing in the electrical product itself; the third is the natural undulating noise. These noises do not meet the requirements of subsequent storage and processing. At this time, it is necessary to pre-process it to eliminate the interference and noise, so as to suppress the clutter independent of the actual signal, and improve the processing ability and accuracy of the subsequent image.
Keywords: image preprocessing; method; accuracy
1 幾種圖像預處理的方法
通常的,圖像的預處理分為圖像的復原和圖像的增強,圖像增強突出了圖像的細節(jié)變化,但同時也放大了圖像的噪聲干擾,圖像的復原降低了噪聲干擾的同時也弱化了圖像的細節(jié)變化[1]。較為理想的圖像預處理方法應該既能消除噪聲干擾,又最大程度地使圖像邊緣輪廓等細節(jié)保持原樣。本研究是對放電光斑進行精確處理,因此對圖像細節(jié)的要求較高??紤]先采用傳統(tǒng)濾波方法對圖像進行預處理,常見的濾波方法有中值平滑濾波、低通濾波、維納濾波等[2]。
由表可見,本研究中采用的方法MSE最低,表示處理后的圖片最接近沒有噪聲的圖片,表明還原能力最好;PSNR最高,表示圖像有用信號和噪聲的比值最大,表明降噪效果最好。
4 結(jié)束語
本文根據(jù)紫外放電圖像的成像特性和本研究中需要對圖像的處理要求,首先對紫外放電圖像進行去噪。先分析了幾種常用的濾波降噪方法,最終提出一種小波域內(nèi)維納濾波的降噪方法。先對噪聲圖像進行小波變換,再利用維納濾波對圖像去噪,并通過在降噪前后取對數(shù)、指數(shù)的形式,有效實現(xiàn)降低紫外放電圖像中的各種噪聲。最后,比較了小波維納濾波與其他各種濾波方法對圖像的處理效果,并使用MSE和PSNR評價處理效果。實驗證明,本研究中使用的濾波法能更為有效地在降低噪聲干擾的同時又最大程度地使圖像邊緣輪廓等細節(jié)保持了原樣。
參考文獻:
[1]金良海.彩色圖像濾波與基于四元數(shù)的彩色圖像處理方法[D].華中科技大學,2008.
[2]吳建華,李遲生,周衛(wèi)星.中值濾波與均值濾波的去噪性能比較[J].南昌大學學報(工科版),1998,01:33-36+62.
[3]崔曉杰.維納濾波的應用研究[D].長安大學,2006.
[4]張德豐,張葡青.維納濾波圖像恢復的理論分析與實現(xiàn)[J].中山大學學報(自然科學版),2006,06:44-47.
[5]衣曉飛,陳福接,楊學軍.基于維納濾波的圖像模板匹配[J].計算機研究與發(fā)展,2000,12:1499-1503.
[6]David G. Lowe. Distinctive image features from scale-invariant key-points[J].The International Journal of Computer Vision, 2004, 60(2): 91-110.
[7]M. Brown, D. G. Lowe. Invariant features from interest point groups[C]. British Machine Vision Conference, 2002.656-665.
[8]Zhang Xin-yang; Zhang Ren-jin, "The technology research in decomposition and reconstruction of image based on two-dimensional wavelet transform," Fuzzy Systems and Knowledge Discovery (FSKD), 2012 9th International Conference on , vol., no., pp.1998,2000, 29-31 May 2012.
[9]李波.數(shù)字圖像噪聲消除算法研究[D].曲阜師范大學,2008.
[10]張旭升,周桃庚,沙定國.數(shù)字圖像噪聲估計的方法及數(shù)學模型[J].光學技術(shù),2005,05:719-722.endprint