摘要
针对传统SUSAN角点提取算法中阈值选取的不确定性,提出了一种自适应分割阈值的SUSAN改进算法。首先采用SUSAN模板对图像进行模板计算得到梯度图,然后根据梯度图的灰度分布特征,采用图像分割方法的判断分析法和KSW熵方法对梯度图做分析处理,最终实现阈值的自动选取,正确提取出有价值的特征角点。试验结果表明,改进算法较之传统算法有明显优势,能准确有效地提取出角点,具有较强的适应性和应用价值。
Aiming at the threshold of the traditional SUSAN point extracting algorithm is uncertain,this paper proposes a modified SUSAN corner detection algorithms based on adaptive threshold.Firstly,the gradient image could be obtained by SUSAN template used in image.Then according to the characteristic of the gradient image gray distribution,using the Judgement analysis mothod and KSW entropy method deal with the gradient image,the threshold automatically is adopted,it can extract valuable feature comer.Using the ASTER remote sensing data as the processing data,we test the traditional algorithm and the modified mothod.Experimental result indicates that the proposed algorithm is superior than the traditional algorithm,which can extract feature points exactly and effectively,and it has more adaptability and applicable value.
出处
《遥感信息》
CSCD
2011年第6期23-27,共5页
Remote Sensing Information
基金
三峡库区地质灾害教育部重点实验室开放基金
863计划项目(2008AA121103)
中国地质调查局地调项目(科[2011]05-02-03)资助
