基于几何特征的图像感兴趣区域的自动定位研究  

Study on Automatic Location for Interested Region of Image Based on Geometrical Features

作　　者：唐毅[1] 郑丽敏[1] 任发政 朱虹[1] 林喆[1] 

机构地区：[1]中国农业大学信息与电气工程学院,北京100083 [2]食品科学与营养工程学院,北京100083

出　　处：《计算机工程》2007年第1期200-203,共4页Computer Engineering

基　　金：国家"863"计划基金资助项目(2002AA248051-2)

摘　　要：自动定位猪肉的臀中肌肉区域是实现基于计算机视觉技术的猪肉无损定级系统的基础,该文针对猪肉的臀中肌肉区域的定位,提出了一种基于几何特征的定位方法,实现了臀中肌区域的自动定位。并与传统的模板匹配法的定位效果进行了比较,实验表明对于定位猪肉的臀中肌基于几何特征的定位法优于传统的模板匹配法,达到了较好的定位效果。Automatic locating the region of gluteus medius is crucial for implementing the lossless grading system for pigs based on computer vision. This paper presents an improved locating method based on geometrical features to locate the region of gluteus medius automatically. And a comparison with the conventional template match method has been performed, experimental results show the presented method is better than template match method to locate the region of gluteus medius and has a perfect effect.

关 键 词：特征提取 模板匹配 目标定位 几何特征 

分 类 号：TP391.41[自动化与计算机技术—计算机应用技术]