基于颜色和形状特征的目标检测与识别  

Object Detection and Recognition Based on Color and Shape Features

作　　者：杜艳红[1] 高庆吉[2] 孙承琦[3] 曾浩[1] 

机构地区：[1]东北电力学院自动化系,吉林吉林132012 [2]哈尔滨工业大学计算机科学与技术学院,黑龙江哈尔滨150001 [3]中国民用航空学院交通工程学院,天津300300

出　　处：《东北电力学院学报》2005年第1期48-53,共6页Journal of Northeast China Institute of Electric Power Engineering

摘　　要：研究了一种具有相似颜色和形状特征的目标检测与识别方法。采用学习 -扩充法获得适应性目标颜色特征的空间分布 ,克服环境光照变化对目标颜色特征提取的影响。利用改进的Hough变换检测目标形状特征 ,解决目标区域出现阴影及目标被部分遮挡对识别的影响。最后通过融合前面所获得的颜色信度和形状信度 ,对目标进行模糊判决。通过移动机器人在复杂环境中对目标的检测与识别应用 。A method of similar objects recognition in complex environment based on color and shape features is proposed. The distributing color space of certain object is obtained by learning-extending method in order to deal with the image captured under the varied lightness. Shape features of certain object with shadow on it or visualizing partly are detected by improved Hough transform. An object is fuzzy judged based on fusing its color belief and shape belief. The validity of this method is proved by the experiments on mobile robot.

关 键 词：形状特征 颜色特征提取 移动机器人 目标检测 识别方法 目标区域 HOUGH变换 影响 有效性 标的 

分 类 号：TM925[电气工程—电力电子与电力传动] TP391[自动化与计算机技术—计算机应用技术]