针对传统的角点检测方法速度较慢、精度不高的情况,为提高速度抑制噪声,提出了一种快速、高精度的角点检测算法。算法在设计上考虑的是USAN区域所对应的弧的像元灰度与角点的相似性来完成角点的提取,使得处理的数据大为减少,同时能保证检测的准确性等其他指标;并且针对阈值k和t固定选取的问题,采用了一种自动选取阈值的方法。通过该算法和SUSAN算法、CSS算法在正确性、漏检、精度等方面的综合比较,结果表明该算法无论对模拟图像还是真实图像均具有良好的性能。

In view of that the traditional image corner detection algorithm accuracy is not high and the speed is low, we propose a fast and novel corner detection algorithm with high precision. The difference between our algorithm and the traditional ones is that the USAN area corresponding to the arc and the corner pixel gray similarity to complete the extraction corner are taken into consideration. Thus, the computational cost is largely reduced and the detection precision as well as other performances are well guaranteed. Considering the issue of the fixed threshold, we adopt a new method that can select the threshold automatically. The proposed method is compared with the SUSAN and CSS corner detectors in accuracy rate, missing detection and precision, and so on. The results show that our algorithm has good performance for both synthesized and real images.