基于几何特征的快速高精度角点检测算法

角点检测应用十分广泛,是许多计算机视觉任务的基础。本文提出了一种快速、高精度的角点检测算法,算法简单新颖,角点条件和角点响应函数设计独特。和以往不同的是:算法在设计上考虑的是角点的局部几何特征,使得处理的数据量大为减少,同时能够很好地保证检测精度等其他性能指标。通过和广泛使用的SUSAN算法、Harris算法在正确率、漏检、精度、抗噪声、计算复杂度等方面进行综合比较,结果表明该算法无论对人工合成图像还是对自然图像均具有良好的性能。

Corner detection is the foundation of many computer vision tasks and widely used. In this paper, we propose a fast and novel corner detection algorithm with high-precision by designing a new corner qualification and a corner response function. The difference between our algorithm and former ones is that local geometric features of the corner are taken into consideration. Thus, computational cost is largely reduced and the detection precision as well as other performances are well guaranteed. The proposed method is compared with SUSAN and Harris corner detectors in accuracy rate, missing detection, precision, anti-noise and computational complexity; the results show that our algorithm has a good performance for both synthesized and real images.

国家自然科学基金(60871078,60835004)资助项目