一种用于形状描述的拱高半径复函数

摘要

A novel contour function, Arc-Height Radius Complex function (AHRC), is proposed for shape description. AHRC captures the global feature and local detail information of the shape through extracting the centroid distance and signed archeight features s from the shape contour. The Fourier coefficients of AHRC are utilized to construct the feature vector for shape description. The performance of the proposed method is evaluated on the standard MPEG-7 shape database and the leaf shape database which is captured from the real world. The experimental results of image retrieval show that the proposed method outperforms the other Fourier descriptors including centroid distance,triangle area,farther-point distance,angle radius and wavelet Fourier descriptors.%提出了一种新的用于形状描述的轮廓线函数-拱高半径复函数(AHRC).AHRC用中心距离和带正负号的拱高来分别描述形状的全局特征和局部细节.用AHRC的傅立叶变换系数构成描述形状的特征向量.在MPEG-7标准测试集上对该方法进行图像检索实验,并将其实际应用于植物叶片图像的检索,同现有的分别基于中心距离、三角形面积、最远点距离、角度半径的傅立叶描述子和小波傅立叶描述子进行性能比较.实验结果表明,本文提出的方法性能最优,表明了该方法的有效性和实际应用价值.