本文選題：Retinex + 圖像增強(qiáng)　； 參考：《哈爾濱工程大學(xué)》2016年碩士論文

【摘要】：圖像增強(qiáng)是圖像處理的重要環(huán)節(jié),其目的對(duì)圖像進(jìn)行加工,按照特定的要求改進(jìn)圖像視覺效果或者增強(qiáng)其中一類信息,突出感興趣的信息或抑制不感興趣的信息,得到一個(gè)效果更好的增強(qiáng)圖像。其中,降質(zhì)圖像增強(qiáng)應(yīng)用最為廣泛,而由于惡劣環(huán)境導(dǎo)致的降質(zhì)圖像由于其成因的復(fù)雜性,特殊性,目前關(guān)于降質(zhì)圖像增強(qiáng)的研究都具有一定的局限性,亟需得到改進(jìn)和加強(qiáng)。因此,本文針對(duì)惡劣環(huán)境,包括天氣、光照、噪聲等原因引起的降質(zhì)圖像的機(jī)理,基于變分框架的Retinex算法,進(jìn)行降質(zhì)圖像增強(qiáng)的研究。本文首先介紹了圖像增強(qiáng)的原理,并對(duì)其中基于Retinex算法的圖像增強(qiáng)方法進(jìn)行了比較,分析其優(yōu)缺點(diǎn)。以基于變分框架的Retinex算法為基礎(chǔ)進(jìn)行研究。針對(duì)其對(duì)反射分量的先驗(yàn)知識(shí)運(yùn)用不足,提出了使用稀疏表示理論改進(jìn)的方法。稀疏表示理論中K-SVD算法對(duì)分段定常性質(zhì)還原性較好的,但是由于字典更新運(yùn)算量過大導(dǎo)致運(yùn)算速度緩慢,因此采用批量正交匹配算法,在字典更新過程中只針對(duì)更新列進(jìn)行計(jì)算,大大減少運(yùn)算量并提升了運(yùn)算速度。并將這種稀疏表示理論應(yīng)用于變分框架的Retinex算法中,對(duì)降質(zhì)圖像進(jìn)行增強(qiáng)�；谏显V理論,使用本文算法增強(qiáng)了針對(duì)不同類型的惡劣環(huán)境造成的降質(zhì)的圖像,為證明算法有效性將仿真結(jié)果與其他算法進(jìn)行了比較,證明了本文算法在亮度提升,對(duì)比度增強(qiáng),結(jié)構(gòu)和細(xì)節(jié)還原上有良好的效果。
[Abstract]:Image enhancement is an important part of image processing. Its purpose is to process the image, to improve the visual effect of the image or to enhance one kind of information according to the specific requirements, to highlight the information of interest or to suppress the information of no interest. Get a better enhancement image. Among them, degraded image enhancement is the most widely used. However, due to the complexity and particularity of the cause of formation, the research on degraded image enhancement has some limitations, which needs to be improved and strengthened. Therefore, aiming at the mechanism of degraded image caused by bad environment, including weather, illumination, noise and so on, this paper studies the enhancement of degraded image based on Retinex algorithm of variational framework. In this paper, the principle of image enhancement is introduced, and the image enhancement methods based on Retinex algorithm are compared, and their advantages and disadvantages are analyzed. The Retinex algorithm based on variational framework is studied. Aiming at the shortage of prior knowledge of reflection component, an improved method using sparse representation theory is proposed. In sparse representation theory, K-SVD algorithm is more effective in reducing piecewise invariability, but because of the slow operation speed due to too much operation of dictionary updating, batch orthogonal matching algorithm is adopted. In the process of dictionary updating, only the update column is calculated, which greatly reduces the computation cost and improves the operation speed. The sparse representation theory is applied to the Retinex algorithm of the variational frame to enhance the degraded image. Based on the appeal theory, this paper uses the algorithm to enhance the degraded image caused by different types of bad environment. In order to prove the effectiveness of the algorithm, the simulation results are compared with other algorithms, and it is proved that the brightness of the algorithm in this paper is improved. Contrast enhancement, structure and detail restoration has a good effect.
【學(xué)位授予單位】：哈爾濱工程大學(xué)
【學(xué)位級(jí)別】：碩士
【學(xué)位授予年份】：2016
【分類號(hào)】：TP391.41

【參考文獻(xiàn)】

相關(guān)期刊論文 前10條

1 王昆;;一種基于變換域的圖像壓縮改進(jìn)算法[J];微型電腦應(yīng)用;2015年11期

2 李倩茹;宋志杰;王良;楊晴;;無方位模糊的非均勻稀疏陣MVDR測向方法[J];聲學(xué)技術(shù);2015年05期

3 楊永波;陳君蕓;;一種超限像素平滑和小波軟閾值圖像去噪算法[J];城市勘測;2015年03期

4 李鑫楠;黃賀艷;賈小寧;馬駟良;;基于指導(dǎo)濾波的圖像盲復(fù)原算法[J];物理學(xué)報(bào);2015年13期

5 張康;;基于空間域低通濾波算子的其他頻譜特性算子獲得方法研究[J];科技廣場;2013年09期

6 陳文飛;許雪峰;苗作華;李康順;;亮度自適應(yīng)的保熵直方圖均衡化方法[J];計(jì)算機(jī)工程與應(yīng)用;2012年16期

7 馬艷娥;陳思;陳娟;劉妍;姚金杰;;基于多幅圖像平均法對(duì)數(shù)字圖像降噪的研究[J];電子測試;2011年06期

8 夏廷君;朱東柏;司昌健;;定參數(shù)Butterworth濾波器的改進(jìn)設(shè)計(jì)[J];黑龍江水專學(xué)報(bào);2010年02期

9 鄭光威;何焰藍(lán);黃水花;譚吉春;王逍;王曉東;;透射型體相位光柵對(duì)連續(xù)激光束的空間低通濾波[J];光學(xué)學(xué)報(bào);2009年04期

10 胡維友;;顏色空間RGB與CMYK之間的變換[J];印刷世界;2007年03期

，

本文編號(hào)：1781707