基于字典学习的超分辨率图像重构.doc
《基于字典学习的超分辨率图像重构.doc》由会员分享,可在线阅读,更多相关《基于字典学习的超分辨率图像重构.doc(101页珍藏版)》请在沃文网上搜索。
1、学位论文独创性(或创新性)声明秉承学校严谨的学风和优良的科学道德,本人声明所呈交的论文是我个人在 导师指导下进行的研究工作及取得的研究成果。尽我所知,除了文中特别加以标 注和致谢中所罗列的内容以外,论文中不包含其他人已经发表或撰写过的研究成 果;也不包含为获得西安电子科技大学或其$教育机构的学位或证书而使用过的 材料。与我一同工作的同志对本研究所做的任何贡献均已在论文中做了明确的说 明并表示了谢意。申请学位论文与资料若有不实之处,本人承担一切的法律责任。本人签名:日期西安电子科技大学关于论文使用授权的说明本人完全了解西安电子科技大学有关保留和使用学位论文的规定,即:研究 生在校攻读学位期间论文
2、工作的知识产权单位属西安电子科技大学。学校有权保 留送交论文的复印件,允许査阅和借阅论文;学校可以公布论文的全部或部分内 容,可以允许采用影印、缩印或其它复制手段保存论文。同时本人保证,毕业后 结合学位论文研究课题再撰写的文章一律署名单位为西安电子科技大学。(保密的论文在解密后遵守此规定)本学位论文属于保密,在_年解密后适用本授权书。本人签名:日期导师签名:日期:I摘要超分辨率图像重构可以看作是一个从单幅或多幅低分辨率图像中重构出一幅 髙分辨率图像的逆问题,近年来被广泛的应用到了视频监控、卫星图像、视频标 准转换、医疗数字影像等各个方面。基于模型和基于学习的方法是最近几年重构 超分辨率图像的两
3、种重要方法。基于模型的方法试图构建低分辨图像到高分辨图 像的映射,其重构效率较高,但由于图像类型多种多样,很难统一到一个模型下 进行描述,在高放大因子下重构图像的质量下降较快。基于学习的方法构造一组 低分辨样例图像和对应的高分辨样例图像,先将待重构图像在低分辨样例图像下 进行编码,再用编码系数来恢复高分辨图像,可以克服基于模型的方法对图像关 系描述不准确的缺陷,具有重构准确、对噪声和图像类型鲁棒性强的等优点。在 基于学习的框架下,本文引入字典学习的方法实现编码,研究了基于字典学习的 超分辨率图像重构方法,所做主要工作如下:(1) 利用KSVD算法学习图像关系。该算法利用K-SVD算法从大量的低
4、分辨率和 高分辨率训练样例图像块中分别训练两个小规模的稀疏字典,利用字典间的对应 关系和低分辨图像块编码来恢复高分辨图像块,不仅能获得更加准确的编码,而 且显著降低了编码的复杂度。(2) 基于多任务字典学习的超分辨率图像重构方法。考虑待重构图像的差异,将训 练样例图像块自组织聚类来训练获得多个字典,用多个字典下的重构来构建多个 任务,不同任务的重构同时进行并共享信息,利用智能的传递特性将多个任务的 结果传递到一个新任务上。多任务算法考虑了训练样例图像块的差异,因此可以 进一步提高单任务的重构质量,而且对含噪图像具有鲁棒性。(3) 基于多任务字典学习和局部约束的超分辨率图像重构方法。假设每个重构
5、图像 块在局部邻域内满足局部结构的相似性,在多任务字典学习的代价函数中加入局 部约束的惩罚项,对重构高分辨图像块进行局部范围内的约束优化,使重构图像 更好的保持结构信息,提高了重构图像的质量。(4) 基于多任务字典学习和残差补偿的超分辨率图像重构方法。对多任务字典学习 重构的图像再进行残差补偿,进一少增强重构髙分辨图像的细节,使重构图像更 好的保持细节信息,提高了重构图像的质量。本文的工作得到了国家自然科学基金(61072108,60601029,60971112)和中央高 校基本科研业务费(JY10000902041)的资助。关键词:超分辨率字典学习多任务学习局部约束残差补偿摘要AIVAbs
6、tractAbstractSuper-resolution image reconstruction (SRIR) is cast as the inverse problem of recovering the original high-resolution (HR) image from one or more low-resolution (LR) images. Recently SRIR has been used in many practical field including medical imaging, satellite imaging and video appli
7、cations and so on. The model-based and the learning-based approach are two most popular SRIR methods that developed in recent years. The model-based approach is of high efficiency; however, the relationships between images are too complex to be expressed under one model. Moreover, when the magnify f
8、actor gets bigger, the reconstructed image degraded quickly. The learning-base approach builds two set of training samples that consists of HR and LR images respectively. The test image is coded under the LR images and the coefficients are taken to recover the HR image using the relationship between
9、 HR and LR training samples. This paper is about the learning-based super-resolution image reconstruction. The main works are as follows:(1) A dictionary learning based SRIR method is proposed. Two dictionaries are learned from the low and high resolution images respectively using K-SVD algorithm. T
10、he proposed algorithm can reconstruct the HR image by making avail of the relationship between the two dictionaries, and it not only have more accurate coding but also significantly reduces the coding complexity.(2) A multitask dictionary learning based SRIR method is proposed. Considering the diffe
11、rences of image blocks,we cluster the training images into several classes from which multiple dictionaries are trained. Single task is defined as recovering the HR image from a dictionary, and the multitask recovery is adopted which share the information among different tasks. For considering the d
12、ifference of the training samples, the proposed method has an improvement on the PSNR of the reconstructed images over the single task counterpart.(3) A local constraint and multi-task dictionary learning based SRIR method is proposed. Local constraint about the structural self-similarity of patches
13、 is added into the cost function. Therefore, it can balance between the maintaining of the local details and the global approximation. Some experiments on natural images results show that it can improve the quality and visual effects of the image.(4) A residual compensation and multi-task dictionary
14、 learning based SRIR method is proposed to optimize the HR image. By adding the residual compensation to the reconstructed image, the proposed method can further refine the details of the edge of the reconstructed image and finally improve the quality and visual effects of the image.This paper was s
15、upported by National Science Foundation of China under Grant no.61072108,60601029,60971112 and the Basic Science Research Fund in Xidian University under Grant no.JY10000902041.Key words: Super-Resolution Dictionary Learning Multitask Learning Local Constraints Residual Compensation第一章绪论M IAbstractI
16、ll 11.1研究背景和意义:11.2研究现状21.3研究内容和创新31.4论文架构安排4第二章超分辨率图像重构模型72.1超分辨率图像重构数学模型72.2超分辨率图像重构步骤92.3传统超分辨率图像重构方法102.3.1非均匀插值算法102.3.2迭代反投影法112.3.3最大后验概率法和最大似然估计法112.4基于学习的超分辨率图像重构方法122.5重构图像质量评价指标132.6本章小结15第三章基于字典学习的超分辨率图像重构173.1图像的稀疏表示173.2基于稀疏表示的超分辨率图像重构193.3KSVD字典学习算法223,4多任务字典学习超分辨率图像重构253.4.1单任务学习算法25
17、3.4.2多任务学习算法263.5实验结果与分析293.5.1自然图像实验结果与分析303.5.2遥感图像实验结果与分析393.6本章小结45第四章基于局部约束的多任务字典学习超分辨率图像重构474.1结构相似性约束滤波474.2超分辨率图像重构中的局部约束484.3基于局部约束和多任务字典学习的超分辨率图像重构504.4实验结果与分析514.4.1自然图像实验结果与分析514.4.2遥感图像实验结果与分析554.5本章小结61第五章基于残差补偿的多任务字典学习超分辨率图像重构635.1残差补偿635.2基于残差补偿和多任务字典学习的超分辨率图像重构645.3实验结果与分析665.3.1自然图
- 1.请仔细阅读文档,确保文档完整性,对于不预览、不比对内容而直接下载带来的问题本站不予受理。
- 2.下载的文档,不会出现我们的网址水印。
- 3、该文档所得收入(下载+内容+预览)归上传者、原创作者;如果您是本文档原作者,请点此认领!既往收益都归您。
下载文档到电脑,查找使用更方便
20 积分
下载 | 加入VIP,下载更划算! |
- 配套讲稿:
如PPT文件的首页显示word图标,表示该PPT已包含配套word讲稿。双击word图标可打开word文档。
- 特殊限制:
部分文档作品中含有的国旗、国徽等图片,仅作为作品整体效果示例展示,禁止商用。设计者仅对作品中独创性部分享有著作权。
- 关 键 词:
- 基于 字典 学习 分辨率 图像