【作者】 贾超； 翟渊博； 张树壮；

【Author】 JIA Chao ZHAI Yuanbo ZHANG Shuzhuang Department of Computer Science and Engineering, Yanshan University, Qinhuangdao Hebei provience, 066004

【机构】 燕山大学信息科学与工程学院；

【摘要】 针对依靠以往的小波方法得到的梯度图来提取边缘将会产生噪声边和坏边的问题,提出了对无位移梯度图利用Contextual滤波器进行边缘检测和多尺度边缘跟踪的方法。通过与其他的方法进行定性和定量的比较,结果表明在对梯度图的优化和进行边缘检测方面要优于以往的基于小波的边缘检测方法。更多还原

【Abstract】 To relying on it is the traditional wave gradient picture where method gets draw edge produce noise of and the bad-edge question. Have proposed utilizing contextual-filter edge detector to measure the edge in there is shift-invariant gradient images, and multiscale edge tracker to solve the problems. Through determining the nature compared with the quantitative one and other methods, indicate the method of this text should be better than prevenient detection methods of edge based on wave transformation.更多还原