發(fā)布時(shí)間：2018-07-06 07:21

  本文選題：顯著性目標(biāo)分割 + 圖像分割��； 參考：《深圳大學(xué)》2016年博士論文

【摘要】：視覺感知在人類獲取周圍環(huán)境的知識(shí)中起到了不可替代的作用,而圖像是視覺感知的重要載體之一。隨著智能手機(jī)的廣泛普及,互聯(lián)網(wǎng)上的圖像資源呈現(xiàn)爆炸性的增長(zhǎng)。如果可以模擬人類的視覺注意感知機(jī)制,對(duì)進(jìn)入視野的海量圖片信息自動(dòng)找出最重要、最顯著的部分進(jìn)行處理,這樣有助于突破信息處理的瓶頸,節(jié)省計(jì)算和分析資源,提高信息處理的速度。顯著性目標(biāo)就是一個(gè)場(chǎng)景中最重要、最吸引人注意的部分。顯著性目標(biāo)分割就是自動(dòng)將輸入圖片中最顯著的目標(biāo)進(jìn)行精確分割的一種模型。近年來,顯著性目標(biāo)分割模型吸引了研究者極大的興趣,涌現(xiàn)出了眾多的優(yōu)秀的模型與算法。而且顯著性目標(biāo)分割技術(shù)在機(jī)器人視覺、視頻安全監(jiān)控、目標(biāo)探測(cè)與識(shí)別、視頻壓縮與編碼、運(yùn)動(dòng)檢測(cè)、視頻目標(biāo)跟蹤等領(lǐng)域得到了廣泛應(yīng)用。本文從基于區(qū)域的圖像分割研究出發(fā),逐步深入到顯著性目標(biāo)分割問題,結(jié)合子空間聚類理論、非局部各向異性擴(kuò)散方程、流擴(kuò)散等相關(guān)理論,提出了一些新的算法和模型。本文工作主要包括以下幾個(gè)方面：1.在基于區(qū)域的圖像分割問題研究中,提出了一個(gè)新的基于非凸半范數(shù)約束的稀疏低秩耦合的圖像分割模型。將子空間聚類模型引入到圖像分割問題中,并且采用S1/2范數(shù)替代核范數(shù)來近似低秩表示,采用L1/2范數(shù)替代L1范數(shù)來近似稀疏表示。研究發(fā)現(xiàn),對(duì)低秩表示問題和稀疏表示問題而言,S1/2范數(shù)和L1/2范數(shù)雖然是非凸的,但它們是一種更好的松弛,且它們的極小化問題具有閉式解。通過ALM算法分離變量,利用半閾值算子得到一種有效的離散算法。大量的實(shí)驗(yàn)證明了模型分割的準(zhǔn)確性和對(duì)噪聲的魯棒性。2.進(jìn)一步深入研究了非凸低秩子空間聚類問題。用Schatten-q擬范數(shù)松弛低秩子空間聚類模型中的秩函數(shù),提出了基于Schatten-q正則化的低秩子空間聚類模型。我們證明了Schatten-q正則最小化問題的最優(yōu)解可以通過一個(gè)廣義的矩陣軟閾值收縮算子表示,模型通過自適應(yīng)線性交替方向乘子法(LADMAP)求解。理論分析證明,廣泛應(yīng)用的軟閾值(SVT)算法和半閾值算法分別是我們新模型在q=1和q=0.5的兩個(gè)特例。大量的實(shí)驗(yàn)證明了我們方法的有效性。3.本文進(jìn)一步研究了顯著性目標(biāo)分割問題。本文將視覺顯著性目標(biāo)檢測(cè)過程建模為一個(gè)非局部各向異性擴(kuò)散方程,提出了一個(gè)基于非局部各向異性擴(kuò)散方程的兩階段顯著性檢測(cè)模型。在第一階段,利用邊界先驗(yàn)進(jìn)行了由邊界到全圖像的背景擴(kuò)散,通過排序的方法選擇了部分顯著性值高的元素作為初始顯著性種子。在第二階段,從顯著性種子出發(fā)將顯著性值擴(kuò)散到全圖像域。在第二階段的顯著性迭代擴(kuò)散過程中對(duì)顯著性種子作了更新和優(yōu)化,消除了錯(cuò)誤種子的的影響,從而有效抑制了背景。對(duì)于模型的求解,本文提出了多方向離散格式來求解新模型。實(shí)驗(yàn)結(jié)果表明,無論從主觀分析,還是從客觀評(píng)價(jià),新方法在3個(gè)常用數(shù)據(jù)庫(kù)上都得到了良好的檢測(cè)效果。4.通過對(duì)現(xiàn)有的基于擴(kuò)散思想的顯著性模型分析發(fā)現(xiàn),現(xiàn)有模型只考慮了點(diǎn)與點(diǎn)之間的擴(kuò)散強(qiáng)度,而沒有考慮到擴(kuò)散方向。這會(huì)導(dǎo)致,兩個(gè)區(qū)域如果特征相似而所處的類別不同時(shí),模型檢測(cè)會(huì)出現(xiàn)較大的誤差。基于這一觀察,本文對(duì)前一個(gè)模型進(jìn)行了改進(jìn),引入了非局部擴(kuò)散張量,以在控制擴(kuò)散強(qiáng)度的同時(shí)考慮擴(kuò)散方向問題。在主方向上,抑制擴(kuò)散從而保證了背景和前景之間的差異性；在其它方向上,新模型提升了擴(kuò)散,從而保證相似部分作為一個(gè)整體高亮檢測(cè)出來。大量實(shí)驗(yàn)證明了新模型的良好性能。5.進(jìn)一步,本文利用多特征的并行框架來模擬人類視覺感知機(jī)制,提出一個(gè)新的基于多特征融合的顯著性流模型。本文提出一個(gè)基本共識(shí),即顯著性目標(biāo)分割時(shí),在大部分特征通道中都顯著的區(qū)域才認(rèn)為是顯著性目標(biāo)�；�于這個(gè)共識(shí),在每個(gè)特征通道中,采用顯著性流擴(kuò)散的方法進(jìn)行建模。然后用二次規(guī)劃的方法訓(xùn)練以得到融合權(quán)重。以調(diào)整過的權(quán)重進(jìn)行線性融合來得到最終的檢測(cè)結(jié)果。經(jīng)過大量的實(shí)驗(yàn)證明我們的模型有很好的綜合性能。
[Abstract]:Visual perception plays an irreplaceable role in the knowledge of human access to the surrounding environment, and image is one of the important carriers of visual perception. With the widespread popularity of smart phones, the image resources on the Internet are explosively growing. If we can simulate human visual perception mechanism, the mass picture letter that enters the field of vision Interest automatically finds the most important and most significant part of the process, which helps to break through the bottleneck of information processing, save computing and analysis resources, and improve the speed of information processing. The significant target is the most important and most attractive part of a scene. The significant target segmentation is the most significant target in the input of the picture. A model for accurate segmentation. In recent years, the significant target segmentation model has attracted the great interest of the researchers, and many excellent models and algorithms have emerged. Moreover, the significant target segmentation technology is in robot vision, video security monitoring, target detection and recognition, video compression and coding, motion detection, video target tracking. Based on the study of region based image segmentation, this paper goes deep into the problem of significant target segmentation, and proposes some new algorithms and modes combining the subspace clustering theory, non local anisotropic diffusion equation, flow diffusion and other related theories. This paper mainly includes the following aspects: 1. In the study of region based image segmentation, a new image segmentation model based on sparse and low rank coupling based on non convex semi norm constraints is proposed. Subspace clustering model is introduced into the image segmentation problem, and S1/2 norm is used instead of kernel norm to approximate low rank representation, and L1/2 norm is used instead of L1 norm to approximate sparse representation. It is found that, for the problem of low rank representation and sparse representation, the S1/2 norm and the L1/2 norm are not convex, but they are a better relaxation, and their minimization problem has a closed solution. A valid discrete algorithm is obtained by using the semi threshold operator by the ALM algorithm. A large number of experiments prove the model points. The accuracy of cutting and the robustness to noise.2. further studies the problem of non convex and low rank subspace clustering. Using the rank function in the Schatten-q quasi norm relaxation low rank subspace clustering model, a low rank subspace clustering model based on the Schatten-q regularization is proposed. We prove the optimal solution of the Schatten-q regular minimization problem. By means of a generalized matrix soft threshold contraction operator, the model is solved by the adaptive linear alternating direction multiplier (LADMAP) method. The theoretical analysis shows that the widely used soft threshold (SVT) algorithm and the semi threshold algorithm are two special examples of our new model in q=1 and q=0.5. A large number of experiments prove the effectiveness of our method.3. This paper further studies the problem of significant target segmentation. In this paper, the visual saliency target detection process is modeled as a nonlocal anisotropic diffusion equation, and a two stage salience detection model based on nonlocal anisotropic diffusion equation is proposed. In the first stage, the boundary prior to the full image is carried out with the boundary prior. In the second stage, the value of saliency is spread to the whole image domain. In the second stage of the significant iterative diffusion process, the significant seeds are updated and optimized to eliminate the effect of the wrong seeds. For the solution of the model, a multi direction discrete scheme is proposed to solve the new model. The experimental results show that the new method, both from the subjective analysis and the objective evaluation, has got a good detection effect on 3 common databases through the analysis of the existing significant model based on the existing diffusion thought. It is found that the existing model only takes into account the diffusion strength between point and point, but does not take into account the diffusion direction. This will lead to a larger error in the model detection in the two regions if the characteristics are similar and the categories are different. Based on this observation, the previous model is improved and the non local diffusion tensor is introduced to the model. The diffusion strength is controlled and the diffusion direction is considered at the same time. In the main direction, the diffusion is suppressed to ensure the difference between the background and the foreground; in other directions, the new model improves the diffusion, thus ensuring the similar part as a whole bright detection. A large number of experiments prove that the good performance of the new model is further.5., Wen Li uses multi feature parallel framework to simulate human visual perception mechanism, and proposes a new model of saliency flow based on multi feature fusion. This paper proposes a basic consensus that significant regions in most characteristic channels are considered explicit targets when significant targets are segmented. Based on this consensus, each feature is connected. In the channel, the model is modeled using the method of significant flow diffusion. Then the fusion weight is trained by two programming methods. The final results are obtained by the linear fusion of the adjusted weights. A lot of experiments show that our model has good comprehensive performance.
【學(xué)位授予單位】：深圳大學(xué)
【學(xué)位級(jí)別】：博士
【學(xué)位授予年份】：2016
【分類號(hào)】：TP391.41

【相似文獻(xiàn)】

相關(guān)期刊論文 前10條

1 詹天明;肖亮;張軍;韋志輝;;基于時(shí)空連續(xù)約束的4D腦圖像分割模型[J];電子學(xué)報(bào);2013年08期

2 董建磊;郝礦榮;李允明;丁永生;;一種新的知覺分割模型[J];計(jì)算機(jī)應(yīng)用研究;2007年04期

3 喻罡;李鵬;繆亞林;卞正中;;多分辨超聲心動(dòng)圖像分割模型[J];西安交通大學(xué)學(xué)報(bào);2006年04期

4 孫曉歡;楊豐;趙祺陽(yáng);;改進(jìn)的分級(jí)多相圖像分割模型及其快速實(shí)現(xiàn)[J];數(shù)據(jù)采集與處理;2010年01期

5 俞璐;謝鈞;吳樂南;;格點(diǎn)上的一種統(tǒng)計(jì)分割模型(英文)[J];Journal of Southeast University(English Edition);2008年01期

6 王海軍;張圣燕;;融合局部和全局信息的兩相腦圖分割模型[J];濱州學(xué)院學(xué)報(bào);2013年06期

7 吳繼明;龐雄文;;一種具有去噪能力的全局最小值分割模型[J];計(jì)算機(jī)工程;2012年07期

8 趙在新;成禮智;;模糊分段光滑圖像分割模型及其快速算法[J];光電子.激光;2011年06期

9 楊曉藝;王小歡;宋錦萍;;連續(xù)最大流圖像分割模型及算法[J];中國(guó)圖象圖形學(xué)報(bào);2013年11期

10 賴軍臣;湯秀娟;謝瑞芝;白中英;李少昆;;基于G-MRF模型的玉米葉斑病害圖像的分割[J];中國(guó)農(nóng)業(yè)科學(xué);2010年07期

相關(guān)會(huì)議論文 前5條

1 徐濤;;時(shí)間序列優(yōu)化分割模型和提取算法[A];2006中國(guó)控制與決策學(xué)術(shù)年會(huì)論文集[C];2006年

2 葉娜;鄭妍;朱靖波;張斌;;基于二維動(dòng)態(tài)規(guī)劃的文本分割模型[A];第三屆全國(guó)信息檢索與內(nèi)容安全學(xué)術(shù)會(huì)議論文集[C];2007年

3 文和平;綦耀光;柯映林;程耀東;;常用刀具的優(yōu)化分割模型[A];第一屆全國(guó)幾何設(shè)計(jì)與計(jì)算學(xué)術(shù)會(huì)議論文集[C];2002年

4 葉娜;羅海濤;鄭妍;朱靖波;張斌;;基于改進(jìn)型Dotplotting的文本分割模型[A];中文信息處理前沿進(jìn)展——中國(guó)中文信息學(xué)會(huì)二十五周年學(xué)術(shù)會(huì)議論文集[C];2006年

5 全剛;徐鐘;孫即祥;;一種改進(jìn)的C-V圖像分割模型[A];第十四屆全國(guó)信號(hào)處理學(xué)術(shù)年會(huì)(CCSP-2009)論文集[C];2009年

相關(guān)博士學(xué)位論文 前4條

1 張秀君;顯著性目標(biāo)分割模型及其應(yīng)用[D];深圳大學(xué);2016年

2 原野;偏微分方程圖像分割模型研究[D];重慶大學(xué);2012年

3 葛琦;基于局部統(tǒng)計(jì)和結(jié)構(gòu)的活動(dòng)輪廓分割模型[D];南京理工大學(xué);2013年

4 章菲倩;數(shù)字幾何模型聯(lián)合分割方法研究[D];南京大學(xué);2015年

相關(guān)碩士學(xué)位論文 前10條

1 李春;基于Chan-Vese圖像分割模型的研究與應(yīng)用[D];中央民族大學(xué);2015年

2 劉金彥;噪聲魯棒的偏微分方程圖像分割模型[D];重慶大學(xué);2015年

3 林靜;PDE分割模型中的局部窗口動(dòng)態(tài)調(diào)整及Bregman實(shí)現(xiàn)[D];福州大學(xué);2014年

4 龔海晏;基于Ambrosio-Tortorelli圖像分割模型的研究與應(yīng)用[D];安徽工業(yè)大學(xué);2016年

5 尹溶芳;兩個(gè)變分圖像分割模型的數(shù)值求解[D];重慶大學(xué);2016年

6 王昊;基于位置與形狀建模的視覺媒體語(yǔ)義分割研究[D];南京大學(xué);2014年

7 張娜;基于全局凸分割模型的兩相圖像分割研究[D];青島大學(xué);2010年

8 王小歡;連續(xù)最大流圖像分割模型及其算法[D];河南大學(xué);2013年

9 彭小朋;一種腦MRI圖像的混合分割模型[D];上海交通大學(xué);2010年

10 李祁鳳;多相M-S模型最小圖分割理論與醫(yī)學(xué)應(yīng)用研究[D];湖南大學(xué);2011年

，

本文編號(hào)：2101949