【摘要】 针对基于多格算法的MIC角点检测方法易混淆边缘点和角点、遗漏角点及角点定位不够精确等缺点,在采用高分辨率的图像后,对低分辨率图像中的候选角点周围的2×2的像素点进行进一步计算,明显地减少了原算法由低分辨率到高分辨率的变化过程中角点的漏检和虚报的概率,同时用梯度的概念进行自适应插值,通过它对角点的候选点作进一步的筛选。测试结果表明提出的方法比改进前的MIC多格算法具有更好的效果。更多还原

【Abstract】 This paper presents an improved approach to corner extraction, which can overcome the MultiGrid algorithm抯 disadvantages of tending to confuse edge points with corners and miss the corners and being unable to determine precisely the position of corners. As the candidate corners in the low-resolution image are further calculated around 2×2 in the high resolution one, the probability of missing corners and false corners is further decreased. The candidate corners, meanwhile, are further extracted based on the self-adapting interpolation with the property of gradient. Experimental results show that the proposed algorithm has a better performance than the original one.更多还原