桥梁拉索表面缺陷的分布式机器视觉检测  

The inspection method based on distributed machine vision for surface defects of bridge cables

作　　者：李新科[1,2] 高潮[1] 郭永彩[1] 贺付亮[1] 

机构地区：[1]重庆大学光电技术及系统教育部重点实验室,重庆400030 [2]重庆大学通信工程学院,重庆400030

出　　处：《光学技术》2013年第5期424-428,共5页Optical Technique

基　　金：教育部博士点基金项目(200806110016);重庆市科委科技计划攻关项目(cstc2012gg-kp1A40005)

摘　　要：介绍了一种桥梁拉索表面损伤缺陷的分布式视觉检测系统。系统以DSP(Digital Signal Processor)芯片TMS320DM642(简称DM642)为核心处理器,协同控制辅助机械装置,实现拉索表面缺陷检测。提出并采用改进的中值滤波对CCD采集到的拉索表面图像进行图像预处理,采用改进的Sobel边缘检测算法对缺陷进行边缘检测;对分割后的缺陷目标进行决策判别,实现桥梁拉索表面缺陷的智能检测。实验结果表明,系统可检测出桥梁拉索表面损伤面积大于1cm2的缺陷,处理速度达到了5cm/s,满足了快速有效的检测要求。A bridge cable surface defects detection system based on distributed machine vision is presentecl, l ne de tection system includes image acquisition, preprocessing, detection and result storage. Preprocessing is conducted on the cable surface images taken by CCDs. Based on the characteristic of the image background noise and the requirement for real time processing, improved median filter image de-noising algorithm is proposed. Improved Sobel edge detection algo rithm is adopted to detect the edge of defects in the images. Defects in the extracted image are detected by area size thresholding. With the same detection accuracy, the complexity of the proposed algorithm is reduced greatly. The system is implemented on a dedicated video processing DSP chip TMS320DM642, and the proposed algorithm successfully ex tracts defects larger than lcm2 in the 510X 576 pixels that the algorithm has good performance in speed and image within O. 932s. Analysis of experiment results demonstrates accuracy.

关 键 词：机器视觉 DSP 桥梁拉索 表面缺陷检测 

分 类 号：TP391.41[自动化与计算机技术—计算机应用技术]