基于图像信息熵的自适应Harris角点检测算法

IMPROVED HARRIS CORNER DETECTION ALGORITHM BASED ON AUTO-ADAPTIVE THRESHOLD ON IMAGE 'S ENTROPY

为了克服Harris角点检测算法在提取角点时需要人为设定响应函数参数和阈值的问题,本文提出一种基于图像信息熵的自适应角点检测算法,利用合适尺度下的图像信息熵能够反映较多图像纹理信息的特点,在非极大值抑制时采用了图像信息熵的自适应阈值,并修改了角点响应函数,避免了阈值设定和参数设定的随机性。通过对添加盐粒噪声的图像进行角点检测并计算角点重复率,表明本文的算法在提取图片角点时有更好的稳定性,能够用于角点检测。

In order to avoid the threshold need to be specified by human in the comer detection of the image, an improved approach on auto- adaptive threshold on image~ entropy to detect comers is proposed in this note. As we know the image~ entropy with right mask can reflect the primary image, using it, an auto - adaptive threshold is adopted in the non - maximum inhibition, and the comer response function is modified to avoid the randomicity of the parameter. The tests show that the improved algorithm has higher superiori- ty and can be used in comer detection.