摘要
为解决传统方法对存在背景噪声干扰的光斑图像分割效果不理想的问题,设计了一种基于梯度卷积的光斑图像分割算法。首先对光斑、背景、噪声的空间灰度分布特性进行分析。而后根据目标与干扰在空间灰度梯度分布特性的差别,构造适应目标尺度的梯度卷积模板,与光斑的梯度图进行卷积,从而在增强光斑特性的同时抑制了背景干扰及噪声。最后,在噪声抑制图像的基础上结合阈值分割和目标特性判别得到最终的分割结果。通过与传统光斑分割算法进行对比实验发现,该算法抗噪能力更强,有效解决了背景与噪声干扰下光斑目标的分割问题。
In order to solve the problem that the traditional method is not ideal for the image segmentation under the noise and background interference,an image enhancement and segmentation algorithm based on gradient image convolution is designed.Firstly,the spatial gray distribution characteristics of the spot,background and noise are analyzed.Then,according to the difference between the target and the interference in the spatial distribution of gradient feature,a convolution template adapted to the target scale is constructed.The template is convolved with the gradient map of the spot image to suppress background and noise while enhancing spot feature.Finally,based on the noise suppression image,the segmentation result of laser spot is obtained by using threshold segmentation and spot feature discrimination.Compared with the traditional spot segmentation algorithm,the experimental results show that under background and noise interference the algorithm has stronger anti-noise ability and effectively solves the problem of laser spot targets segmentation.
作者
杨世坤
杨逸峰
姜丽辉
左乐
YANG Shi-kun;YANG Yi-feng;JIANG Li-hui;ZUO Le(Shanghai Aerospace Control Technology Institute,Shanghai 201108,China)
出处
《激光与红外》
CAS
CSCD
北大核心
2020年第6期754-760,共7页
Laser & Infrared
关键词
光斑图像
图像分割
梯度
卷积
降噪
背景噪声
spot image
image segmentation
gradient
convolution
noise reduction
background noise
