Research And Design On High-speed Press Quality Image Detection System

The traditional method for detecting the quality of print has many disadvantages, suchas low efficiency, high rejection rate. This thesis researches and develops themachine-vision-based high-speed image detection system. The thesis conducts the researchof proactive detection and passive detection. The static image detection system need s theparticipation of operator, and it belongs to the proactive detection system. But the thesisadopts the SIFT algorithm to detecting the image, and the whole process is conducted by theIPC. And this system belongs to the passive detection system.This thesis first introduces the general principal of press quality image detection basedon machine vision, and discusses its application significance and prospect. According to thedemands of the machine vision and print detection, the thesis designs the hardwarecomponents of the detection system based on machine vision. And the thesis designs thesystem’s software to complete the function of the detection system. Since all the operationsof the setup of the system parameters and control actions are completed by the specialcontrol keyboard, the software also needs to deal with the virtual key code. Due to thereal-time requirements of the image detection system, this thesis adopts the multi-threadedprogramming techniques. The multi-threaded programming technique improves the resourceutilization and throughput, gives full play to the processors’ and peripherals’ ability to workin parallel, and improves the program concurrency. In the part of image processing anddisplaying, the thesis adopts the DirectDraw multimedia programming interfaces to make theoperations faster, stronger and more effective. In addition, the thesis uses the OpenCVcomputer vision library and the SIFT algorithm to extract the detected press image features,and makes the compares to the standard images. The thesis introduces the concepts ofmatching degree, recall and precision rate. And by optimizing both the precision and therecall rate to set the best matching degree threshold, the system can achieve the automatic detection of the press quality of images.The industrial applications and experiments show that the system play a very importantrole in the press quality image detection, and this system has important application value.