针对SUSAN角点算法中计算过程复杂及其使用固定阈值的局限性,提出一种基于提升小波变换的自适应改进方法.该算法根据角点的分布特性,引入快速提升小波变换理论,在图像的高频区域筛选出候选角点,缩小需要精确检测的角点范围,提高了算法效率;并根据图像局部灰度信息自动调节核心点与其邻域像素的灰度差值,代替原算法中的单一阈值,以提高算法自适应能力.实验结果证明了该方法的快速有效性.
Aiming at the limitations of using fixed threshold and the complex calculation process in SUSAN operator. An adaptive SUSAN comer detector based on lifting wavelet transform is proposed in the paper. According to the comer properties in an image, the theory of lifting wavelet transform is adopted to process the image. It will be more effective that the comer candidates are obtained from the high frequency information, which can reduce the searching range for comers The new algorithm is self adaptive to adjust the intensity discrepancy between the core point and its epsilon neighborhood pixels. Experiment demonstrates that the proposed comer detector is faster and more effective than traditional SUSAN algorithm.