Design Of Multi-camera Real-time Cluster Acquisition And Processing System

With the rapid development of modern industry, the use of increasingly high cost of artificial, enterprises must improve the level of automation in order to improve their competitiveness, aiming at large-scale parallel processing of image data is studied.This article is a high-speed parallel processing of multi-camera system to study, usually in the detection of the liquid crystal panel, glass, circuit boards and other aspects of wider application. In this paper, the application of the liquid crystal panel as an example of the experimental procedure. Industrial production is usually relatively large size LCD panels, in order to realize the entire surface of the liquid crystal panel defect detection, industrial applications requiring multiple cameras simultaneously on the LCD panel surface with high precision image acquisition, resulting in an amount of image data acquired Very big. In order to improve the image processing speed, the goal of this paper is the use of real-time processing system MPI cluster parallel computing and related communication algorithm design for a multi-camera acquisition.This paper is divided into three parts:the first part first introduced the research background of the subject, and then gives the four kinds of common system scheme design through to these four schemes were analyzed and summarized, the final choice of a distributed cluster based on MPI processing system as the experimental platform of the system structure. The second part introduces the development of cluster communication system. The development of communication system includes remote boot, remote shutdown, TCP communication between nodes, multi process and data communication and so on. Part three introduces the MPI parallel programming implementation and whole experiment system operation process and use logging tool to record the running process of the system, through the log file analysis to understand the performance of the whole system.