【摘要】 对最小亮度变化角点检测进行了深入探讨,对其存在的不足提出两点改进:⑴用模糊度概念进行自适应窗口插值,大大减少了以往最小亮度变化角点检测普遍采用固定窗口插值所造成的漏检和虚报概率;⑵利用斜线上边缘点的方向连续性,有效地滤除了由于图像量化误差而造成的伪角点.实验证明,改进的算法明显地提高了角点检测概率和准确度.更多还原

【Abstract】 On the basis of studying MIC corner detection, two improved methods are suggested. First, the concept of fuzziness is used to put forward a new adaptive window interpolation algorithm to MIC corner detection. This algorithm is superior to the widely adopted algorithm with fixed window interpolation algorithm which will cause the missing of the right corner or false corner alarming. Second, making use of the characteristic that the edge pixels in the oblique line have continuity in direction, some kind of false corners caused by the quantificaiton error are removed easily. Experimental results show that the probability and accuracy of corner detection are improved greatly.更多还原