Research Of The Image Processing Method On The Structure Light Vision Measurement

As society continues to progress,people's demand for knowledge of the world is growing,the traditional machine vision technology can not meet the demands for3-dimensional object,and so three-dimensional vision measurement technology has become a hot research.Structured light 3-D visual inspection has been used more and more widely in modern industry with its own characteristics,such as strong real-time performance,high precision and speed.In practical applications,all kinds of geometric shapes of the measured objects and surface information,while changing the shooting mode,the venue and light.These issues make the structured light image not be directly used subsequent 3-D reconstruction.Therefore,processing the structure of the optical image is a key step in the process of light visual inspection.This paper makes the deep research on the structured light image filtering,enhancement and segmentation.The specific work of this paper is as follows:(1)Due to the surrounding environment impacting to light(natural light or illumination light),it makes the structured light images with common noise,thus it degrades the quality of the image.In order to improve the median filter that can not filter out the noise and accelerate the algorithm,this paper further study an improved adaptive median filtering.The experiments show that this improved filter algorithm can acquire better effect of de-noising and improve the speed,it is very effective not only for ordinary image,but also for structured light image.(2)The brightness of the structured light image is uneven because of lighting problems,or the detail of the image is lost due to a lack of exposure.Based on the two problems,this paper research a kind of improved multi-scale Retinex enhancement algorithm based on the HSV color space.The experiments show that this algorithm can not only enhance the highlights of the details,but also can improve the details in the shadows,at the same time,it also implements the color fidelity,which makes good preparation for subsequent processing.(3)After image filtering and enhancement to get the structured light image,it also contains background information and other irrelevant information,which is not unfavorable for subsequent processing.This paper further study a fast two-dimensional maximum entropy threshold segmentation method on Sobel operator.This method not only greatly lowers the computational complexity,also improve the efficiency,and it resolves the problem of the superimposed image edges.The experiments show that the algorithm issuitable to ordinary images and structured light images.