本文選題：小波框架 + 運(yùn)動(dòng)圖像修復(fù)�。� 參考：《湘潭大學(xué)》2015年碩士論文

【摘要】：在數(shù)字圖像領(lǐng)域由于運(yùn)動(dòng)造成的降質(zhì)圖像的修復(fù)問題是一個(gè)挑戰(zhàn)性問題。在現(xiàn)實(shí)生活中隨著交通工具和電子設(shè)備的日漸發(fā)達(dá),對(duì)違反交通規(guī)則的車輛由于高速行駛被拍下多幅運(yùn)動(dòng)模糊的車牌號(hào)圖像,或者是刑事案件中的犯罪嫌疑人在匆忙經(jīng)過路邊攝像頭時(shí)被拍下的多幅運(yùn)動(dòng)模糊圖像等這些普遍存在、運(yùn)動(dòng)復(fù)雜而又非常重要的圖像進(jìn)行修復(fù)有著非常重要的現(xiàn)實(shí)意義。這些情況的原始圖像和模糊核都未知,修復(fù)難度相對(duì)較大,屬于多幅盲運(yùn)動(dòng)圖像修復(fù)。由于利用多幅盲運(yùn)動(dòng)模糊圖像進(jìn)行圖像修復(fù)時(shí),可以獲得較大的圖像信息量,能夠處理運(yùn)動(dòng)復(fù)雜的圖像修復(fù)問題,并能夠獲得較好的修復(fù)效果。因此,多幅盲運(yùn)動(dòng)圖像的修復(fù)算法具有很重要的研究意義。在圖像修復(fù)領(lǐng)域由于緊小波框架的冗余性能夠?qū)е卖敯粜?使得在低精度下獲得的圖像分解系數(shù)能夠重構(gòu)出高精度的清晰圖像,在圖像修復(fù)方面有著很高的重構(gòu)精度,因此對(duì)緊小波框架在圖像修復(fù)領(lǐng)域的應(yīng)用有著很大的研究價(jià)值。本文結(jié)合緊小波框架的優(yōu)勢和圖像分類處理的思想提出基于緊小波框架的多幅盲運(yùn)動(dòng)分類圖像修復(fù)方法,該方法對(duì)紋理圖像和分段光滑圖像分別建立圖像修復(fù)模型,同時(shí)在修復(fù)模型中考慮了圖像中的噪聲,加入了噪聲約束項(xiàng)。實(shí)驗(yàn)表明,該方法既能提高不同類型的運(yùn)動(dòng)圖像修復(fù)質(zhì)量,又能有效地去除運(yùn)動(dòng)圖像中的噪聲。另外,由于本文提出的基于緊小波框架的多幅盲運(yùn)動(dòng)分類圖像修復(fù)方法中的緊小波框架是固定的,對(duì)于不同圖像,當(dāng)其在進(jìn)行緊小波框架分解時(shí)不能得到合適的稀疏度,進(jìn)而影響圖像的修復(fù)質(zhì)量,因此,在后期的研究中本文又提出多幅盲運(yùn)動(dòng)分類圖像的自適應(yīng)緊小波框架修復(fù)方法,該方法在交替迭代過程中,利用奇異值分解(Single Value Decomposition,SVD)對(duì)用來構(gòu)成緊小波框架分解矩陣的濾波器組進(jìn)行求解,獲得適應(yīng)于圖像的緊小波框架分解矩陣,并將該框架分解矩陣應(yīng)用于多幅盲運(yùn)動(dòng)分類圖像修復(fù)中。實(shí)驗(yàn)表明該方法的圖像修復(fù)效果有所提高。
[Abstract]:In the field of digital images, the problem of restoration of degraded images caused by motion is a challenging problem. In real life, with the increasing development of vehicles and electronic devices, vehicles that violate traffic regulations are photographed with a number of motion-blurred license plates due to high speed driving. It is very important to repair the motion blur images that the suspects in criminal cases take when they pass the roadside camera in a hurry. These images which are complex and very important have very important practical significance. The original image and blur kernel of these cases are unknown, so the restoration is relatively difficult and belongs to multiple blind moving images. Because of the large amount of image information can be obtained when many blind motion blur images are used, it can deal with the problem of image restoration with complex motion, and can obtain a better effect. Therefore, it is of great significance to study the restoration algorithm of multiple blind moving images. In the field of image restoration, the redundancy of compact wavelet frame can lead to robustness, so that the image decomposition coefficient obtained under low precision can reconstruct high precision clear image, and has high reconstruction accuracy in image restoration. Therefore, the application of compact wavelet framework in the field of image restoration has great research value. Combined with the advantages of compact wavelet framework and the idea of image classification, this paper proposes a new image restoration method based on compact wavelet framework for blind motion classification, which establishes image restoration models for texture images and segmented smooth images, respectively. At the same time, the noise in the image is taken into account in the restoration model, and the noise constraint term is added. Experiments show that the proposed method can not only improve the quality of different kinds of moving images, but also effectively remove the noise from moving images. In addition, because the compact wavelet framework proposed in this paper is fixed in the blind motion classification image restoration method based on compact wavelet framework, for different images, when it is decomposed by compact wavelet frame, it can not get the appropriate sparse degree. Therefore, in the later stage of the research, the adaptive compact wavelet frame restoration method for multiple blind motion classification images is proposed, which is used in the process of alternating iteration. Using singular value decomposition (SVD) single value Decomposition (SVD), the filter banks used to form a compact wavelet frame factorization matrix are solved, and a compact wavelet frame factorization matrix suitable for images is obtained. The frame decomposition matrix is applied to the restoration of several blind motion classification images. Experiments show that the image restoration effect of this method is improved.
【學(xué)位授予單位】：湘潭大學(xué)
【學(xué)位級(jí)別】：碩士
【學(xué)位授予年份】：2015
【分類號(hào)】：TP391.41;O174.2

【相似文獻(xiàn)】

相關(guān)期刊論文 前9條

1 劉冬青;;巧用運(yùn)動(dòng)圖像解題[J];考試周刊;2011年61期

2 霍洪峰;趙煥彬;;運(yùn)動(dòng)圖像解析標(biāo)定框架改進(jìn)及算法[J];實(shí)驗(yàn)室研究與探索;2014年02期

3 方勇;張瑾;;一種基于傅里葉變換的運(yùn)動(dòng)圖像全盲認(rèn)證方法[J];應(yīng)用科學(xué)學(xué)報(bào);2007年03期

4 曲巨寶;林宏基;梁洪濤;劉勝;;運(yùn)動(dòng)圖像快速跟蹤技術(shù)研究[J];重慶師范大學(xué)學(xué)報(bào)(自然科學(xué)版);2011年01期

5 賀建峰,肖楠楠,陳鋼;運(yùn)動(dòng)圖像的模糊與恢復(fù)[J];昆明理工大學(xué)學(xué)報(bào);1997年02期

6 劉旭東;;小差異運(yùn)動(dòng)圖像中關(guān)鍵興趣姿態(tài)點(diǎn)的差異特征分離[J];科技通報(bào);2014年02期

7 李騰達(dá);趙曉煒;;隨機(jī)體育運(yùn)動(dòng)圖像捕捉點(diǎn)虛擬重建中凌亂誤差控制[J];科技通報(bào);2014年02期

8 馬生元;勢箱中電子運(yùn)動(dòng)圖像的繪制[J];青海大學(xué)學(xué)報(bào)(自然科學(xué)版);2002年02期

9 ;[J];;年期

相關(guān)會(huì)議論文 前3條

1 李前進(jìn);李林;趙煥彬;張有會(huì);蘭連意;劉紅娟;;運(yùn)動(dòng)圖像解析中的三維重構(gòu)[A];第八屆全國體育科學(xué)大會(huì)論文摘要匯編（二）[C];2007年

2 訾方;趙大煒;張科;李言俊;;寬帶自然運(yùn)動(dòng)圖像的速度估計(jì)[A];2006中國控制與決策學(xué)術(shù)年會(huì)論文集[C];2006年

3 李玉剛;柏開祥;嚴(yán)發(fā)本;;運(yùn)動(dòng)圖像解析系統(tǒng)圖像采集丟幀與系統(tǒng)精度的分析[A];第十一屆全國運(yùn)動(dòng)生物力學(xué)學(xué)術(shù)交流大會(huì)論文匯編（摘要）[C];2006年

相關(guān)重要報(bào)紙文章 前7條

1 劉琳;運(yùn)動(dòng)圖像領(lǐng)域?qū)＠治鯷N];中國知識(shí)產(chǎn)權(quán)報(bào);2013年

2 本報(bào)記者　楊谷;用手機(jī)看電影無須等3G[N];光明日?qǐng)?bào);2002年

3 劉全恩;降耗技術(shù)獲突破等離子應(yīng)用將更廣泛[N];中國電子報(bào);2008年

4 笨笨火狐;用Real Producer Plus批量壓縮影片[N];電腦報(bào);2001年

5 電子工業(yè)部四所博士后 殷玉U，

本文編號(hào)：2006223