人工智能技术的大视场光电测量系统畸变校正  

Distortion correction of photoelectric measurement system with large field of view based on artificial intelligence

作　　者：柳赟[1] 鹿冠群 LIU Yun;LU Guanqun(School Of Electrical and Electronic Engineering,North China Electric Power University,Beijing 102206,China)

机构地区：[1]华北电力大学电气与电子工程学院,北京102206

出　　处：《激光杂志》2020年第8期40-44,共5页Laser Journal

基　　金：中央高校基本科研业务费专项资金项目(No.2019MS005)。

摘　　要：传统方法在大视场光电测量系统畸变校正过程中,存在成像离散程度较高、像素畸变偏移量较大和校正时间较长的问题,因此,在分析大视场光电测量系统畸变成像的基础上,提出基于人工智能技术的图像畸变校正方法。确定发生畸变的像素点到图像中心的距离,并构建多项式图像校正数学模型在空间范围内对畸变像素点做坐标转换。在此基础上针对分割图像块边界位置的奇异像素点,利用优化支持向量机技术对大视场图像进行全局优化,识别并剔除位于图像块边界处的畸变像素点,进一步降低图像的失真程度,从而实现对大视场光电测量系统畸变的校正。分析结果表明,提出图像畸变校正方法的离散化程度相对于传统方法更低,同时在像素畸变偏移量控制及时间性能方面也具有优势。In the process of distortion correction of large field of view photoelectric measurement system,the traditional method has some problems,such as high imaging discreteness,large pixel distortion offset and long correction time.Therefore,based on the analysis of distortion imaging of large field of view photoelectric measurement system,an image distortion correction method based on artificial intelligence technology is proposed.The distance between the distorted pixels and the image center is determined,and the polynomial image correction mathematical model is constructed to transform the coordinates of distorted pixels in space.On this basis,aiming at the singular pixels in the boundary of segmented image block,the optimized support vector machine technology is used to optimize large field of view image globally,identify and remove the distorted pixels located at the boundary of the image block,further reduce the image distortion,so as to achieve distortion correction of large field of view photoelectric measurement system.The analysis results show that the discretization degree of proposed method is lower than traditional method,and has advantages in pixel distortion offset control and time performance as well.

关 键 词：光电测量 人工智能技术 多项式模型 支持向量机 

分 类 号：TN206[电子电信—物理电子学]