基于机器学习和方向模板的遥感图像边缘检测方法  

Remote Sensing Image Edge Detection Method Based on Machine Learning and Haar Template

作　　者：袁宇丽[1] YUAN Yuli(School of Computer Science,Neijiang Normal University,Neijiang,Sichuan 641100,China)

机构地区：[1]内江师范学院计算机科学学院,四川内江641100

出　　处：《内江师范学院学报》2020年第8期51-55,共5页Journal of Neijiang Normal University

摘　　要：针对遥感图像的边缘检测尤其是彩色遥感图像的边缘检测上存在诸多的不足,为了减少噪声,提高目标提取的精确度与完整性,提出一种基于机器学习和方向模板的遥感图像边缘检测方法.算法首先将RGB彩色图像转换到YCbCr颜色空间,根据机器学习获得的分类阈值进行分割,再采用改进的Harr方向特征模板,获得了较为完整的彩色遥感图像边缘信息.实验表明,基于机器学习算法的边缘检测能较好的提取出遥感图像的边缘特征,特别是对于地面物体的轮廓特征提取较为完整,符合遥感图像目标提取与识别需求.There are too many shortcomings in edge detection of remote sensing image,especially in color remote sensing image.In order to reduce noise and improve the accuracy and integrity of target extraction,a remote sensing image edge detection method based on machine learning and direction template is proposed.Firstly,RGB color image is transformed into YCbCr color space,and then segmentation is done according to the classification threshold value obtained by machine learning.Then,the improved Harr direction feature template is used to obtain more complete edge information of color remote sensing image.The experiment reveals that the edge detection based on machine learning algorithm can extract the edge features of remote sensing images,especially for the contour features of ground objects,which meets the special needs of target extraction and recognition of remote sensing images.

关 键 词：遥感图像 边缘检测 机器学习 方向模板 

分 类 号：TP751[自动化与计算机技术—检测技术与自动化装置]