【摘要】：近些年來,隨著城市中車輛和人口密度不多增加,空氣質(zhì)量遭到嚴(yán)重破壞,導(dǎo)致霧霾現(xiàn)象頻發(fā)。霧霾的存在,致使能見度大大降低,圖像采集設(shè)備無法直接獲得準(zhǔn)確的信息,進(jìn)而對周圍環(huán)境作出錯(cuò)誤的判斷,嚴(yán)重的甚至?xí)䦟?dǎo)致災(zāi)難發(fā)生。霧天情況下,由于空氣中漂浮著大量懸浮顆粒,光線在傳播過程中會(huì)被散射而造成強(qiáng)度衰減,光線強(qiáng)度的衰減使得圖像對比度下降,細(xì)節(jié)模糊,色彩保真度下降。因此,霧霾圖像的清晰化處理具有重要的現(xiàn)實(shí)意義,圖像去霧領(lǐng)域越來越受到國內(nèi)外研究者的關(guān)注。在眾多研究成果中,何愷明博士的暗通道去霧算法給圖像去霧技術(shù)領(lǐng)域帶來了新的啟發(fā),該算法具有簡單有效、實(shí)時(shí)性和自動(dòng)去霧的優(yōu)點(diǎn),其缺點(diǎn)主要是對大氣散射物理模型過度依賴、無法選取相匹配的濾波模板大小、對天空缺乏適用性以及恢復(fù)后圖像偏暗。本文在大氣散射物理模型基礎(chǔ)上,分析了霧天圖像質(zhì)量下降的原因,并針對暗通道去霧算法進(jìn)行實(shí)驗(yàn)、改進(jìn)與嘗試,并取得了較好的成像效果。本論文的工作主要體現(xiàn)在以下幾點(diǎn):(1)提出了一種基于多尺度的思想來細(xì)化透射率的方法,從而達(dá)到去霧效果。該方法以暗通道先驗(yàn)準(zhǔn)則為理論基礎(chǔ),在保證暗通道先驗(yàn)準(zhǔn)則與區(qū)域內(nèi)的透射率相同假設(shè)成立的同時(shí),防止估計(jì)的透射率圖像在暗通道突變的區(qū)域發(fā)生塊狀效應(yīng)。(2)在求取暗通道時(shí)采用自適應(yīng)的方法進(jìn)行濾波,在提升對比度的同時(shí)保持了結(jié)構(gòu)信息。對原始的大氣光值求取過程進(jìn)行改進(jìn),有效抑制了天空區(qū)域去霧后色彩過飽和的現(xiàn)象。(3)對于透射率的估計(jì),為了克服天空區(qū)域復(fù)原之后呈現(xiàn)彩色的光暈問題,本文通過分析透射率特點(diǎn),在獲得初始透射率之前的中間過程進(jìn)行特殊處理,然后對其進(jìn)行引導(dǎo)濾波優(yōu)化,最終獲得了較好的復(fù)原效果。實(shí)驗(yàn)結(jié)果充分表明了本文提出的改進(jìn)方法良好的改進(jìn)效果,且較其他改進(jìn)方法具有簡單并且更有效的特點(diǎn),但是本文的改進(jìn)方法并不能適用各類含霧圖像,尤其是對含有大量天空或明亮區(qū)域的圖像,去霧效果并不理想,值得繼續(xù)深入研究和改進(jìn)。
[Abstract]:In recent years, with the small increase of vehicle and population density in the city, the air quality has been seriously damaged, resulting in the frequent occurrence of haze phenomenon. Because of the existence of haze, the visibility is greatly reduced, and the image acquisition equipment can not directly obtain accurate information, and then make a wrong judgment on the surrounding environment, which will even lead to disaster. In the case of fog, due to the floating of a large number of suspended particles in the air, light will be scattered in the process of propagation, resulting in intensity attenuation, the attenuation of light intensity makes the image contrast decrease, the details are blurred, and the color fidelity decreases. Therefore, the clarity of haze image processing is of great practical significance, and the field of image fog removal has been paid more and more attention by researchers at home and abroad. Among the many research results, Dr. he Kaiming's dark channel fog removal algorithm has brought new inspiration to the field of image defogging technology. the algorithm has the advantages of simple and effective, real-time and automatic fog removal. The main disadvantage of the algorithm is that it is too dependent on the physical model of atmospheric scattering, unable to select the matching filter template size, lack of applicability to the sky and dark image after restoration. In this paper, based on the physical model of atmospheric scattering, the reasons for the decline of image quality in fog days are analyzed, and the dark channel fog removal algorithm is tested, improved and tried, and good imaging results are obtained. The work of this paper is mainly reflected in the following points: (1) A method based on multi-scale idea to refine the transmittance is proposed to achieve the effect of fog removal. Based on the theory of dark channel prior criterion, while ensuring that the dark channel prior criterion is the same as the transmittance assumption in the region, the estimated transmittance image is prevented from block effect in the region of dark channel mutation. (2) the adaptive method is used to filter the dark channel, which not only improves the contrast, but also maintains the structure information. The original atmospheric light value calculation process is improved, which effectively suppresses the phenomenon of color supersaturation after fog removal in the sky region. (3) for the estimation of transmittance, in order to overcome the color halo problem after sky region restoration, this paper analyzes the transmittance characteristics, carries on the special treatment to the intermediate process before obtaining the initial transmittance, and then carries on the guide filter optimization to it, and finally obtains the better restoration effect. The experimental results fully show that the improved method proposed in this paper has a good improvement effect, and has the characteristics of simple and more effective than other improved methods, but the improved method in this paper is not suitable for all kinds of fog images, especially for images with a large number of sky or bright areas, and the fog removal effect is not ideal, which is worthy of further study and improvement.
【學(xué)位授予單位】：蘭州交通大學(xué)
【學(xué)位級別】：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 馬忠麗;文杰;;融合邊緣信息的單尺度Retinex海霧去除算法[J];計(jì)算機(jī)輔助設(shè)計(jì)與圖形學(xué)學(xué)報(bào);2015年02期

2 錢澤東;何勇;;基于改進(jìn)Retinex算法的含霧圖像清晰化處理技術(shù)[J];科技創(chuàng)新導(dǎo)報(bào);2015年03期

3 肖勝筆;李燕;;具有顏色保真性的快速多尺度Retinex去霧算法[J];計(jì)算機(jī)工程與應(yīng)用;2015年06期

4 孫小明;孫俊喜;趙立榮;曹永剛;;暗原色先驗(yàn)單幅圖像去霧改進(jìn)算法[J];中國圖象圖形學(xué)報(bào);2014年03期

5 李利榮;汪蒙;;一種高效的圖像增強(qiáng)去霧算法[J];湖北工業(yè)大學(xué)學(xué)報(bào);2013年05期

6 李沛軒;葉俊勇;;基于小波變換和模糊理論的裂紋圖像增強(qiáng)算法[J];計(jì)算機(jī)系統(tǒng)應(yīng)用;2013年09期

7 王子須;于素萍;;一種避免顏色失真的圖像增強(qiáng)算法[J];計(jì)算機(jī)工程與應(yīng)用;2013年12期

8 蔡超峰;任景英;;基于直方圖均衡化的手背靜脈圖像對比度增強(qiáng)[J];計(jì)算機(jī)應(yīng)用;2013年04期

9 張冰冰;戴聲奎;孫萬源;;基于暗原色先驗(yàn)?zāi)Ｐ偷目焖偃レF算法[J];中國圖象圖形學(xué)報(bào);2013年02期

10 李朝鋒;唐國鳳;吳小俊;琚宜文;;學(xué)習(xí)相位一致特征的無參考圖像質(zhì)量評價(jià)[J];電子與信息學(xué)報(bào);2013年02期

相關(guān)碩士學(xué)位論文 前4條

1 賈冬雪;圖像去霧算法的研究與應(yīng)用[D];沈陽工業(yè)大學(xué);2015年

2 陳永亮;灰度圖像的直方圖均衡化處理研究[D];安徽大學(xué);2014年

3 何小波;面向監(jiān)控場景的去霧方法研究[D];華北電力大學(xué);2014年

4 宋玉婷;基于三維彩色直方圖均衡化的彩色圖像增強(qiáng)算法研究[D];山東財(cái)經(jīng)大學(xué);2013年

，

本文編號(hào)：2505351