基于相邻角点间隔度数直方图图像检索

提出一种新的基于图像形状的图像检索方法，首先将图像分成中心圆和外侧圆环两个区域，每个区域分别求出所有相邻角点的间隔度数，将间隔度数量化若干等级，从而计算出每幅图像在每个区域的相邻角点的间隔度数直方图，两幅图像的相似性取决于对应区域的直方图的欧式距离．该算法具有平移、旋转及尺度不变性的特点，经实验证明具有较好的图像检索效率．

Interval Degree Histogram Image Retrieval Based on the Adjacent Corner

This paper presents a new image retrieval method based on shape, first the image is divided into central and lateral annular two region, adjacent corner interval degree is calculated in the region, the interval is quantified some grades, and thus calculate each image in each region of the adjacent corner interval histogram, the two images of the similarity depends on the corresponding region histogram Euclidean distance. The algorithm has the translation, rotation and scale invariant features, the experiment proves that it has better image retrieval efficiency.