宽边缘细化及改进Hough变换的快速稳像算法  

Fast image stabilization algorithm based on multi-pixel edge thinning and improved Hough transform

作　　者：王斌[1] 赵跃进[1] 

机构地区：[1]北京理工大学信息科学技术学院光电工程系,北京100081

出　　处：《红外与激光工程》2009年第1期166-169,共4页Infrared and Laser Engineering

基　　金：国家部委预研项目(42001070302)

摘　　要：数字稳像是用来解决由于摄像机所处平台不稳定造成的图像抖动问题,而图像本身模糊带来的图像边缘模糊会给稳像带来不利影响。针对视频图像边缘模糊造成的宽边缘问题,提出一种基于数学形态学思想的快速图像边缘细化算法,同时针对Hough变换计算量大、效率低、易受平行线干扰的缺点,提出改进Hough变换的快速运动估计算法。实验表明:此算法舍弃了大部分的冗余量,提高了计算效率,能快速有效地确定补偿矢量;改进算法可重复性较好,提高了算法的稳定性和运算速度,且对运动物体有抗干扰作用。Image jitter problem is caused by unsteady camera flat. Digital image stabilization is used to solve this problem. Fuzzy image can bring bad effect on image stabilization. To solve multi-pixel edge problem caused by fuzzy edge of video image, fast image edge thinning algorithm based on mathematical morphology was proposed. At the same time, rapid motion estimation algorithm was proposed to resolve the defects of Hough transform,such as large calculation, low efficiency and often disturbed by parallellines. Experiment results prove that this method gives up the great mass of redundancy, enhances the calculation efficiency, and can obtain compensated vectors fast and effectively. Repeated experiments show that this algorithm has better repeatability, stability, computation speed, and anti-jamming to moving object.

关 键 词：稳像 形态学梯度 边缘细化 HOUGH变换 

分 类 号：TP391[自动化与计算机技术—计算机应用技术]