發(fā)布時(shí)間：2018-05-11 05:32

  本文選題：大氣散射模型 + 大氣耗散函數(shù)　； 參考：《西安科技大學(xué)》2017年碩士論文

【摘要】：隨著電子電路技術(shù)和光學(xué)成像技術(shù)的發(fā)展,獲取的圖像分辨率、清晰度越來(lái)越高。然而,在霧霾等惡劣天氣條件下,捕捉的圖像會(huì)發(fā)生嚴(yán)重退化,給航空航天、工業(yè)生產(chǎn)、交通監(jiān)管及生物醫(yī)學(xué)等領(lǐng)域帶來(lái)極大的影響。導(dǎo)致圖像降質(zhì)的主要原因是場(chǎng)景目標(biāo)光線在傳播過(guò)程中與空氣中懸浮的顆粒發(fā)生交互作用,從而使得圖像細(xì)節(jié)信息丟失、色調(diào)偏移、飽和度降低,無(wú)法滿足人類的視覺(jué)要求。為了提高戶外成像系統(tǒng)適用性和穩(wěn)定性,本文基于大氣散射模型的圖像去霧算法進(jìn)行仔細(xì)研究,并提出適用性強(qiáng)、穩(wěn)定性高的去霧算法。本文主要研究?jī)?nèi)容如下:針對(duì)傳統(tǒng)偏振去霧算法認(rèn)為偏振度為全局不變量,導(dǎo)致的圖像整體偏暗、層次感低、色調(diào)低沉、丟失圖像細(xì)節(jié)信息,提出了改進(jìn)的基于均值濾波的偏振圖像去霧算法。首先利用偏振成像探測(cè)系統(tǒng)獲得最好與最差狀態(tài)下的兩幅偏振圖,然后通過(guò)四叉樹細(xì)分法估算大氣光強(qiáng),實(shí)現(xiàn)了大氣光強(qiáng)等有關(guān)參數(shù)的自動(dòng)尋優(yōu),并利用均值濾波的變形形式估算出大氣散射光,再根據(jù)去霧模型復(fù)原無(wú)霧圖像,最終實(shí)現(xiàn)圖像去霧。實(shí)驗(yàn)表明該算法場(chǎng)景適應(yīng)力強(qiáng),具有更高的對(duì)比度,更清晰的圖像細(xì)節(jié)和更豐富的色調(diào),有一定的視覺(jué)優(yōu)勢(shì)。針對(duì)基于中值濾波的快速去霧方法存在的不足,提出了一種基于總變差的快速霧天圖像復(fù)原算法。首先利用總變差對(duì)大氣耗散函數(shù)進(jìn)行估計(jì),有效的避免了邊緣殘霧和局部偏暗現(xiàn)象;然后引入直方圖修正機(jī)制下的自適應(yīng)保護(hù)因子,更正明亮區(qū)域的大氣散射函數(shù);同時(shí)運(yùn)用何算法通過(guò)暗通道快速準(zhǔn)確的獲取大氣光值,降低了時(shí)間復(fù)雜度,避免了霧圖中高亮物體對(duì)大氣光值估計(jì)的影響。最后對(duì)復(fù)原結(jié)果進(jìn)行亮度調(diào)整,得到顏色更加豐富,細(xì)節(jié)更加清晰的去霧效果圖,且算法的時(shí)間復(fù)雜度是圖像像素?cái)?shù)的線性函數(shù),在計(jì)算速度上取得了較大的提升。
[Abstract]:With the development of electronic circuit technology and optical imaging technology, the resolution and sharpness of the obtained images are becoming higher and higher. However, under severe weather conditions such as haze, the captured images will degenerate seriously, which will have a great impact on the fields of aerospace, industrial production, traffic regulation and biomedicine. The main cause of image degradation is the interaction between the light of the scene target and the suspended particles in the air during the propagation process, which results in the loss of the details of the image, the color offset and the decrease of the saturation, which can not meet the visual requirements of human beings. In order to improve the applicability and stability of outdoor imaging system, the image de-fogging algorithm based on atmospheric scattering model is studied carefully in this paper, and a suitable and stable de-fogging algorithm is proposed. The main contents of this paper are as follows: according to the traditional polarization de-fogging algorithm, the degree of polarization is considered as a global invariant, which leads to the overall dark image, low hierarchy, low tone, loss of image detail information. An improved de-fogging algorithm for polarization images based on mean filter is proposed. First, two polarizations in the best and worst states are obtained by using the polarization imaging detection system, and then the atmospheric light intensity is estimated by the quadtree subdivision method, which realizes the automatic optimization of the relative parameters, such as atmospheric light intensity. The atmospheric scattering light is estimated by the deformed form of mean filter, and then the fog free image is reconstructed according to the defog model, and finally the image is de-fogged. Experimental results show that the algorithm has better adaptability, higher contrast, clearer image details and richer hue, and has some visual advantages. A fast fogging image restoration algorithm based on total variation is proposed to overcome the shortcomings of the fast de-fogging method based on median filter. Firstly, the total variation is used to estimate the atmospheric dissipation function, which effectively avoids the edge fog and the local dark phenomenon, and then the adaptive protection factor based on histogram correction mechanism is introduced to correct the atmospheric scattering function in the bright region. At the same time, we use the algorithm to get the atmospheric light value quickly and accurately through dark channel, which reduces the time complexity and avoids the influence of the highlight object in the fog map on the atmospheric light value estimation. Finally, the brightness of the restoration results is adjusted to obtain a more colorful, more clear detail de-fogging effect, and the time complexity of the algorithm is a linear function of the number of pixels of the image, and the computational speed has been greatly improved.
【學(xué)位授予單位】：西安科技大學(xué)
【學(xué)位級(jí)別】：碩士
【學(xué)位授予年份】：2017
【分類號(hào)】：TP391.41

【參考文獻(xiàn)】

相關(guān)期刊論文 前10條

1 戴聲奎;;去霧算法中天空區(qū)域直方圖分析與模式識(shí)別[J];信號(hào)處理;2015年12期

2 劉興云;戴聲奎;;消除halo效應(yīng)和色彩失真的去霧算法[J];中國(guó)圖象圖形學(xué)報(bào);2015年11期

3 曾浩;尚媛園;丁輝;周修莊;付小雁;;基于暗原色先驗(yàn)的圖像快速去霧[J];中國(guó)圖象圖形學(xué)報(bào);2015年07期

4 張登銀;鞠銘燁;王雪梅;;一種基于暗通道先驗(yàn)的快速圖像去霧算法[J];電子學(xué)報(bào);2015年07期

5 夏宏麗;李鋼;張仁斌;李興興;;基于偏振特性的圖像去霧算法[J];計(jì)算機(jī)應(yīng)用與軟件;2014年10期

6 王偉鵬;戴聲奎;;結(jié)合圖像融合與分割的快速去霧[J];中國(guó)圖象圖形學(xué)報(bào);2014年08期

7 張小剛;唐美玲;陳華;湯紅忠;;一種結(jié)合雙區(qū)域?yàn)V波和圖像融合的單幅圖像去霧算法[J];自動(dòng)化學(xué)報(bào);2014年08期

8 張冰冰;戴聲奎;孫萬(wàn)源;;基于暗原色先驗(yàn)?zāi)Ｐ偷目焖偃レF算法[J];中國(guó)圖象圖形學(xué)報(bào);2013年02期

9 陳龍;郭寶龍;畢娟;朱娟娟;;基于聯(lián)合雙邊濾波的單幅圖像去霧算法[J];北京郵電大學(xué)學(xué)報(bào);2012年04期

10 禹晶;徐東彬;廖慶敏;;圖像去霧技術(shù)研究進(jìn)展[J];中國(guó)圖象圖形學(xué)報(bào);2011年09期

相關(guān)碩士學(xué)位論文 前2條

1 曾佳;基于大氣散射模型的霧天圖像復(fù)原方法研究[D];合肥工業(yè)大學(xué);2015年

2 王多超;圖像去霧算法及其應(yīng)用研究[D];安徽大學(xué);2010年

，

本文編號(hào)：1872606