低分辨率图像目标特征实时增强方法仿真  

Simulation of Low-Resolution Image Target Feature Real-Time Enhancement Method

作　　者：陶卫平[1] TAO Wei - ping(Qinzhou University, Qinzhou Guangxi 535011 China)

机构地区：[1]钦州学院,广西钦州535011

出　　处：《计算机仿真》2017年第10期208-212,共5页Computer Simulation

基　　金：2015年度广西高等教育本科教学改革工程项目(2015JGA371);广西自治区级精品视频公开课<机械工程专业导论>(桂教高教[2016]62号第33号)

摘　　要：为了提高分辨率较低的目标图像的实时可读性,保护低分辨率图像的边缘和目标特征,需要进行低分辨率图像目标特征实时增强。针对当前图像增强方法通常会导致低分辨率图像边缘和目标的模糊的问题,提出一种基于多尺度top-hat分解的低分辨率图像目标特征实时增强方法。上述方法首先对低原分辨率图像利用多尺度top-hat变换构造塔形分解细节特征和近似图像,再对原低分辨率图像的细节特征应用融合规则,最后根据不同应用需求加权各尺度目标特征获得低分辨率图像目标特征融合图像,采用正则化方法直接对低分辨率图像目标特征进行处理,以附加稀疏性约束的形式,将低分辨率图像目标特征增强问题转化为最优化问题,并采用数值方法进行快速求解,由此完成低分辨率图像目标特征实时增强。实验结果表明,所提方法不仅能够有效抑制低分辨率图像噪声,还能保护并增强图像目标和边缘特征,具有较好的鲁棒性。This research focuses on real - time enhancement method for target characteristic of low - resolution image. Firstly, the research used multi - scale top - hat conversion to build detail feature of pyramid decomposition for the image and carried out fusion rule for detail feature of the original image, then according to different application requirement, weighted target characteristic of each scale to obtain fusion image of the target characteristic. Moreover, the research adopted regularization method to process the target characteristic directly and converted the problem of target characteristic enhancement into optimization problem in form of addition sparse constraint. The research also employed numerical method to carry out fast solution. Experimental results show that the method can not only sup- press noise of the low - resolution image effectively, but also protect and enhance image target and edge feature. It has better robustness.

关 键 词：低分辨率图像 目标特征 实时增强方法 

分 类 号：TP391.4[自动化与计算机技术—计算机应用技术]