基于颜色及投影特征的挖掘机图像分割算法  

Segmentation Algorithm of the Excavator Image Based on Color and Projection Characteristics

作　　者：李源[1] 何荣开[1] 王庆[2,3] 张小国[2,3] 

机构地区：[1]东南大学机械工程学院,南京210096 [2]东南大学仪器科学与工程学院,南京210096 [3]东南大学苏州研究院,江苏苏州215123

出　　处：《小型微型计算机系统》2013年第11期2635-2638,共4页Journal of Chinese Computer Systems

基　　金：国家"十二五"科技支撑计划项目(2012BAJ23B02)资助

摘　　要：通过对挖掘机图像的RGB颜色特征的分析,提出一种基于颜色信息的挖掘机图像分割算法,该算法采用直接二值化进行分割,而不需要进行灰度变换.对分割后的挖掘机图像进行投影分析后,基于其波形特征,提出一种再分割方法,该方法可以有效地分割出挖掘机的机械臂部分,以简化后续的识别过程.通过对分割后图像的各封闭区域的面积测量,将小面积的区域予以消除,能够有效地消除孔洞现象及去除较大的噪声,有利于对机械臂的进一步识别.本文提出的算法为通过机械臂来识别挖掘机提供了预处理算法,处理速度快,分割精度高.After analyzing the RGB color characteristics of excavator image, we propose a segmentation algorithm based on the color information of the excavator image, directly by using binary segmentation without requiring gray-scale transformation. After the pro- jection analysis of the segmentation image, according to its waveform characteristics, we put forward a method to segment the robotic arm portion. Through measuring the area of each enclosed region, the holes and larger noises can be effectively eliminated after remo-ving the small area regions, which is conducive to the further identification. The presented method and algorithm in this paper provide a preprocessing algorithm of excavator recognition by identifying robotic arm, with quick speed and high accuracy.

关 键 词：颜色特征 图像分割 投影波形 面积测量 

分 类 号：TP391[自动化与计算机技术—计算机应用技术]