基于灰度共生矩阵的草地未割区域分界线提取  

Extraction of Border Line of Non-cut-lawn Based on GLCM

作　　者：童逸舟 刘瑜[1] 

机构地区：[1]浙江理工大学机械与自动控制学院,浙江杭州310018

出　　处：《杭州电子科技大学学报（自然科学版）》2016年第2期62-66,71,共6页Journal of Hangzhou Dianzi University：Natural Sciences

摘　　要：利用草地的灰度图像计算灰度共生矩阵,通过比较"自相关"、"对比度"、"能量"、"均匀性"4个不同的特征值矩阵,提取出纹理分割效果最佳的二值图像,然后将二值图横向分割成若干宽度相同的横条,通过计算单条横条内黑色像素点的占有比确定历遍轮廓大小进而找出分割点坐标,最后将每个横条分割点用直线连起来.实验表明,利用灰度共生矩阵能较好地将未割草地的纹理特性提取出来,并通过后期算法处理能相对准确地进行分界线的提取.The extraction of border line of non-cut-lawn will be helpful for the path planning of intelligent lawn mower.This algorithm calculates the gray level co-occurrence matrix（GLCM）through the gray scale image of the lawn.By comparing the four different eigenvalue matrixes＂correlation＂＂contrast＂＂energy＂and＂homogeneity＂,a binary image which has the best effect of texture segmentation can be extracted,and then be divided into a plurality of aequilate transverses.The size of contour traversal can also be determined by calculating the occupancy of black pixels in a single transverse and therefore the coordinate of the breakpoints can be found out.Consequently,these breakpoints from each transverse are connected with straight lines.This experiment shows that by using GLCM the texture characteristics as well as the border line of non-cut-lawn can be accurately extracted.

关 键 词：草地分割 灰度共生矩阵 特征值矩阵 历遍轮廓 

分 类 号：TP391.4[自动化与计算机技术—计算机应用技术]