Abstract: This paper analyzes the theory of Harris operator, and points out its disadvantage. It proposes an auto-adaptive Harris operator based on the ideology of image break and neighboring point eliminating method. It makes the corners are well-proportioned distributing, and avoids too many corners stay together effectively. Experimental results show that the algorithm can detect corner more reasonable, and can be used in image mosaics well.

Key words: corner detection, Harris operator, feature extraction

摘要： 分析Harris角点检测算子的实现原理及其不足，提出一种基于图像分块和邻近角点剔除策略的自适应Harris角点检测算法，检症测出的角点均匀分布，有效避免了角点聚簇现象。实验结果表明，该算法检测出的角点分布更均匀合理，能很好地适应图像拼接等实际 应用。

关键词: 角点检测, Harris算子, 特征提取

CLC Number: 

• TP391