Abstract: This paper proposes a corner detection method based on gray level difference analysis. The processing of the novel corner detector assumes a well-defined corner of gray level difference analysis, which performs effectively on the basis of the adaptive thresholds to local window with less error and failure. With the aid of “gray-level ridge” track and angle-distinguished accumulation principle, not only coordinates but also corner attributes, such as corner angles and corner orientations, are available. Without any processing of edge detection or pre-segmentation for contours, experimental results prove that the proposed approach works well over most images, and is fast enough for real time applications.

Key words: Image processing, Corner detection, Pattern recognition, Gray level difference

摘要： 提出了一种新的基于灰度差统计的角点检测方法。该方法以原始图像全图的灰度差统计结果作为出发点，采用的灰度脊跟踪算法和累积角度判别原则不仅能提供角点的准确位置，还能确定各角点的开张角度和方向。由于省略了边缘检测过程，因此较大地提高了角点检测的效率。实验证明该方法应用在大部分真实图像的角点检测效果优异，同时运行速度能满足大部分角点检测应用的需要。

关键词: 图像处理, 角点检测, 模式识别, 灰度差