Research Of Out-of-focus Image Restoration & Micro-crack Identification System Based On DSP Of The Automatic Magnetic Particle Testing System

The thesis includes mainly two parts: the research of out-of-focus image restoration of the automatic magnetic particle testing system and hardware design of micro-crack image processing system based on DSP.The automatic magnetic particle testing system improves efficiency and reliability of part testing, while it substitutes computer vision for operator's watching. In fact, the images acquired are often blurred by the out-of-focusing, which sets the obstacle for micro-crack identification, farther leads to false identification. In the thesis, the knowledge and principle on magnetic particle testing, digital image and image restoration are introduced first. According to demand of the system, the Wiener filter restoration based on parameter identification is proposed and analyzed by the Matlab experiment, the specification of evaluating effect of image restoration is presented according to out-of-focus radius as well. Programming by Visual C++6.0, JPEG image files are decoded and converted into BMP files, colorful image is converted into gray image and out-of-focus micro-crack image is restored. The restoration technique and VC++ program application resolve the "bottle-neck" problem of the system.Based on the realization of PC's software system, hardware design of micro-crack image processing system based on DSP is studied. The structure and features of TI CO.TMS320VC5410 are introduced, hardware design and debugging are accomplished and experiment data are gained, the distinction between DSP and PC is compared. The hardware design on DSP resolves the problem of the system speed and attains to the expected effect.