Paper: S. Belongie , J. Malik , J. Puzicha, Shape Matching and Object Recognition Using Shape Contexts, IEEE Transactions on Pattern Analysis and Machine Intelligence, v.24 n.4, p.509-522, April 2002
只就其中Shape Context相关核心部分阅读。
问题描述:一些在肉眼看来相似的图形,在旋转、变形等细小变化后就无法被传统图形识别方法(Feature-Based、Brightness-Based)识别。文中提出了一种简单且鲁棒的方法,来寻找图形间的一致性。
文中将图形识别问题分为3步:
1) 解决两个图形间的对齐问题;
2) 估计两者间的对齐变换;
3) 用此对齐变换的matching error计算两个图形间的距离。
对图形识别应用Shape Context方法的步骤大致如下:
1) 获取图形的轮廓边缘,并用离散点表示如(a)(b)。这些点不一定是极点或角点,当点的采样越高越能表现图形的细节;
2) 对于该点集中的每一个点P,应用一个指数-极坐标。该指数-极坐标被分为r个半径区域,theta个角度区域,共N个bin。该坐标系给于靠近参照点的点更高的权重,使得Shape Context方法对局部更敏感;
3) 对每一个bin,统计除点P外的轮廓点的个数,得到如(d)(e)(f)的直方图;
4) 对两个图形上的每一对点,使用一下公式计算它们之间的匹配代价;
5) 则使得所有C值的和最小的匹配为对齐变换。
Shape Context无视图形的旋转、缩放等变形,在这方面类似Hu等不变矩,但较之鲁棒得多。但由于其是基于图形轮廓边缘的算法,在应用于实际图像时,若前景与背景很难区分或提取的轮廓效果不好出现漏洞断裂等情况,则不能达到很好的效果。
此日志的引用通告 URL 是:
http://lincccc.spaces.live.com/blog/cns!CDE197AF0E29D7CE!148.trak
分享到:
相关推荐
shape matching and object recongnition using shape contexts
关于SHAPE CONTEXT 提出及应用的经典文章
经典原文文献shape Matching and object Recognition Using shape的代码,代码有中文注释,易懂
Chapter 2:- OBJECT DETECTION AND RECOGNITION USING FEATURES - Working with Object Recognition - Features and Descriptors - Feature Matching and Homography - Building an Application
Fast object retrieval using direct spatial matching Fast object retrieval using direct spatial matching Fast object retrieval using direct spatial matching
z_A Fast Shape Context Matching Using Indexing_2011
主要讲的事图像处理中的图像匹配,对图像的匹配讲的非常详细。
Matlab demo code for shape context matching with thin plate splines(Serge Belongie, Jitendra Malik and Jan Puzicha)
Object retrieval with large vocabularies and fast spatial matching,经典的关于视觉词袋模型应用的论文
classifier system to a complex letter recognition task. A set of 20,000 unique letter images was generated by randomly distorting pixel images of the 26 uppercase letters from 20 different commercial ...
IbPRIA 2005 (Iberian ...2.Shape and Matching 3.Image and Video Processing 4.Image and Video Coding 5.Face Recognition 6.Human Activity Analysis 7.Surveillance 8.Robotics 9.Hardware Architectures
Stereo Processing by Semiglobal Matching and Mutual Information (SGBM)论文原文
FACE RECOGNITION USING ENSEMBLE STRING MATCHING
Face recognition by fusion of local and global matching scores using ds theory An evaluation with uni-classifier and multi-classifier paradigm,关于人脸识别的一篇不错的文章
基于形状的模板匹配,Edge Based Template Matching Implementing an edge based template matching or pattern matching algorithm.
Semi-Global Matching in Object Space 原文
STRING-MATCHING AND OTHER PRODUCTS
You can get a very accurate amplitude, ... Matching Pursuit and orthogonal matching pursuit, The signal spectral analysis and filtering, Pattern Recognition bayes discriminant analysis algorithm.
Matching with Shape Contexts,介绍如何将形状上下文用于图像匹配
a presentation of a series of computationally efficient algorithms for testing graph isomorphism and related graph matching tasks in pattern recognition and a new graph distance measure to be used ...