基于多尺度曲率乘积角点检测算法的优化与改进

Optimization and Improvement of Corner Detection Algorithm Based on MultiScale Curvature Product

红外图像的角点检测在模式识别、计算机视觉和目标跟踪等领域中占有重要地位,为了提取到更精准的红外图像的角点,提出在基于曲率多尺度积算法(MSCP)的基础上的改进算法,增强提取到的角点。实验表明:该方法检测效率高,提取到的角点更稳定,精度更高,可满足实际需求中的高精准度的需求。

Corner extraction of infrared image plays an important position in the field of pattern recognition, computer vision, and other related occu-pies, in order to better corner detection, proposes the infrared image based on the multi-scale product curvature algorithm（MSCP） on the basis of improvement, to enhance the corner of the detected point. The actual test shows that this method is of high detecting efficiency,extract the corner is more stable, higher precision, can satisfy the actual application of real-time, high precision requirements.