基于三维光学模型的图像边缘检测效果优化  

Optimization of image edge detection based on 3D optical model

作　　者：于晓明[1] 刘广达[1] YU Xiaoming;LIU Guangda(Eastern Liaoning University,School of Mechanical Engineering,Dandong Liaoning 118003,China)

机构地区：[1]辽东学院机械工程学院,辽宁丹东118003

出　　处：《激光杂志》2020年第8期91-95,共5页Laser Journal

基　　金：辽宁省科学技术厅项目(No.20141129)。

摘　　要：传统方法提取图像边缘信息时容易受外界干扰,导致提取特征不完整、检测结果精准度低,因此,提出了一种基于三维光学模型的图像边缘检测效果优化方法,首先分析三维光学模型分析待检测图像的特性,进而获得微型彩色像素内彩色边缘效应图。之后使用入射光与反射光转换法,凭借多尺寸Retimex把图像划分成位置、大小、方向都各不相同的分量,使感兴趣的分量提升从而完成对图像的增强,随后利用迭代法对全局阈值划分处理得到二值图像,最后将直接使用Sobel算子检测的图像边缘与二值图像进行融合,从而完成对目标图像边缘检测。仿真实验证明,所提方法能够增强检测效率与精准度,其中检测精度最高可达98%,检测耗时最少为0.95 s。It’s easy to be interfered by outside when extracting image edge information by traditional methods,resulting in incomplete extraction features and low accuracy of detection.Therefore,this paper proposes an image edge detection effect optimization method based on three-dimensional optical model.Firstly,the three-dimensional optical model is constructed to analyze image characteristics to be tested.Then,the color edge effect image in micro color pixel is obtained.After that,the image is divided into different positions,sizes and directions by using multi-dimensional retimex to enhance the interested components,and then the binary image is obtained by using iterative method to divide global threshold.Finally,the image edge detected by Sobel operator is fused with binary image to complete the edge detection of target image.The simulation results show that this method can enhance detection efficiency and accuracy,in which the detection accuracy is up to 98%,and detection time is at least 0.95 s.

关 键 词：边缘检测 三维光学模型 SOBEL算子 图像划分 

分 类 号：TN02[电子电信—物理电子学]