基于边界和梯度特征的图像识别  

Image Recognition Based on Edge and Gradient Character

作　　者：黄外斌[1,2] 张亶[3] 陈刚[1] 

机构地区：[1]浙江大学计算机图像图形研究所 [2]浙江大学城市学院信息与计算科学系,杭州310015 [3]浙江大学计算机学院

出　　处：《计算机工程》2004年第9期56-57,161,共3页Computer Engineering

基　　金：国家自然科学基金资助项目(60202002);浙江大学城市学院教师科研基金

摘　　要：提出了一种基于边界和梯度特征的图像识别方法。在相差仿射变换的图像识别过程中，将边界点和其对应的梯度特征组合，既使特征具有仿射变换下的不变性，又提高了识别的效率。严格的数学推导证明了算法在理论上的可行性，实验结果表明了该算法在实际应用中取得了较好的效果。In this paper, an image recognition based on edge and gradient character is proposed. In the process of recogning, the image different from affine transformation, the character based on both edge character and gradient character is utilized. This type of character is affine invariant and its stability is improved, in addition it is more efficient to calculate these characters. Mathematic proof represents that this approach is valuable and the simulation shows the result is acceptable.

关 键 词：图像识别 边界 梯度 图像处理 

分 类 号：TP391.41[自动化与计算机技术—计算机应用技术]