摘要提出了一种改进的Har“s角点检测方法。该方法在Harris角点检测求得角点响应函数后，利用双掩膜来定

义进行非极大值抑制的局部范围，结合K均值聚类方法进行非极大值抑制，若像素点的角点响应函数值满足预设角点判定

条件，则将该像素点定义为角点。实验结果表明，该方法无需进行阈值选择，提高了角点检测精度。-Abst旧ct An improved harris_based comer detection methOd is presented in the paper． Firstly，the comer response

function of the Harris comer detection should be obtained Secondly，the tworrmsks are defined the scope of mrr-maxmla

suppression and the Knleans algorithrn is used to start on the non．ma)【ima suppressiorL Thirdly，if the value of the pixel cor—

ner response function is agree with the preset conditions，the pixel is defined as the comer． The experiments show that this

rrlethod does not require the threshold selection and improve the precision of comer detection．

Key WOrcIs Harris comer detection，feature extraction，K meaJls