本文選題：暗通道先驗　切入點：K-means聚類　出處：《青島大學》2017年碩士論文　論文類型：學位論文

【摘要】：霧天海邊環(huán)境中,大氣中的懸浮粒子中水滴含量較高,大氣光的散射作用更強,霧的濃度更大,獲取的圖像質量退化,影響了計算機對圖像的分析、識別與處理。由此可見,對霧天圖像進行有效的去霧處理,增加圖像的清晰度,提高圖像的質量,有著現(xiàn)實和理論的迫切需要。目前的圖像去霧方法的研究,大多是基于暗通道先驗算法理論,圖像去霧效果比較顯著。但是,這種方法仍然存在天空色彩失真、圖像處理速度慢的缺點。本文針對這些不足,結合海邊霧天圖像的特征,提出一種基于暗通道先驗的海邊霧天圖像去霧算法。針對暗通道算法在處理天空區(qū)域時色彩失真的現(xiàn)象,本算法對大氣光強度的估計方法進行了改進。對含天空區(qū)域較多的海邊霧天圖像,運用改進的K-means聚類方法分割原始圖像,得到保留比較完整圖像信息的天空區(qū)域。整幅圖像的大氣光強度估計值是由對分割的圖像分別估計天空區(qū)域和非天空區(qū)域的大氣光強度值并進行加權處理得到。針對軟摳圖算法用時過長問題,本算法采用導向濾波器對透射率進行優(yōu)化,不但縮短時間,還能實現(xiàn)邊緣平滑、增強細節(jié),有效消除方塊效應,得到相對準確的透射率估計值。由于海邊霧天圖像較暗,色彩不明顯,圖像對比度低,對于復原的圖像進行色調重映射,提高了圖像的對比度。本文提出的基于暗通道先驗的海邊霧天圖像去霧算法借助MATLAB實驗平臺進行仿真,并與中值濾波和雙邊濾波兩種算法進行結果比對,實驗證明本算法能有效去除霧天對海邊圖像的影響,使圖像更自然明亮,對比度更高,可視性更好,實現(xiàn)霧天圖像的快速、有效處理。
[Abstract]:In the foggy seaside environment, the droplets in the suspended particles in the atmosphere are higher, the scattering of atmospheric light is stronger, the concentration of fog is greater, the quality of the obtained images is degraded, and the analysis, recognition and processing of the images are affected by the computer. There is a realistic and theoretical urgent need for effective defog processing to increase the sharpness and improve the quality of fog images. Most of the current researches on image de-fogging methods are based on the theory of dark channel priori algorithm. The effect of image de-fogging is remarkable. However, this method still has the disadvantages of color distortion of sky and slow image processing. A priori defog algorithm for seaside fog images based on dark channel is proposed. The color distortion of dark channel algorithm in dealing with the sky region is discussed. The algorithm improves the estimation method of atmospheric light intensity. The improved K-means clustering method is used to segment the original images for seaside fog images with more sky regions. The atmospheric light intensity estimation of the whole image is obtained by estimating the atmospheric light intensity of the segmented image in the sky region and the non-sky region, respectively. Aiming at the problem that soft matting algorithm takes too long time, In this algorithm, the transmissivity is optimized by using a guide filter, which can not only shorten the time, but also realize the edge smoothing, enhance the detail, eliminate the square effect effectively, and get the relatively accurate transmittance estimation value. The color is not obvious and the contrast of the image is low, so the reconstructed image is remapped to improve the contrast of the image. In this paper, the image de-fogging algorithm based on the dark channel priori is simulated with the help of the MATLAB experimental platform. The experimental results show that the algorithm can effectively remove the influence of fog on the seaside image, make the image more natural and bright, the contrast is higher, the visibility is better, and the image of fog can be quickly realized, and the results are compared with the results of median filter and bilateral filtering, and the experimental results show that the proposed algorithm can effectively remove the influence of fog on the seaside image. Effective handling.
【學位授予單位】：青島大學
【學位級別】：碩士
【學位授予年份】：2017
【分類號】：TP391.41

【參考文獻】

相關期刊論文 前10條

1 霍海蒙;徐世許;王漢萍;;基于暗原色先驗的海邊霧天圖像去霧[J];計算機與現(xiàn)代化;2016年07期

2 劉興云;戴聲奎;;消除halo效應和色彩失真的去霧算法[J];中國圖象圖形學報;2015年11期

3 楊燕;白海平;;基于暗通道先驗的補償快速去霧算法[J];計算機工程;2015年10期

4 曾浩;尚媛園;丁輝;周修莊;付小雁;;基于暗原色先驗的圖像快速去霧[J];中國圖象圖形學報;2015年07期

5 李加元;胡慶武;艾明耀;嚴俊;;結合天空識別和暗通道原理的圖像去霧[J];中國圖象圖形學報;2015年04期

6 楚君;王華彬;陶亮;周健;;基于引導濾波器的單幅霧天圖像復原算法[J];計算機工程與應用;2015年21期

7 王偉鵬;戴聲奎;;基于域變換遞歸濾波的霧天圖像復原[J];計算機應用研究;2015年02期

8 劉紅平;陳明義;;基于二次濾波的單幅圖像快速去霧算法研究[J];計算機工程與應用;2015年08期

9 孫小明;孫俊喜;趙立榮;曹永剛;;暗原色先驗單幅圖像去霧改進算法[J];中國圖象圖形學報;2014年03期

10 王一帆;尹傳歷;黃義明;王洪玉;;基于雙邊濾波的圖像去霧[J];中國圖象圖形學報;2014年03期

相關碩士學位論文 前5條

1 彭惠珍;基于暗通道先驗的霧天圖像清晰化方法研究[D];中國海洋大學;2015年

2 肖杰;基于暗通道先驗的霧霾天交通場景下退化圖像復原算法研究[D];西南交通大學;2015年

3 董林娜;基于暗原色先驗的圖像去霧算法研究[D];山東師范大學;2015年

4 許允棟;K-means聚類算法的改進與應用[D];廣西師范大學;2015年

5 劉佳婕;惡劣環(huán)境下圖像去霧算法的研究及實現(xiàn)[D];南京郵電大學;2015年

，

本文編號：1641136