Study On Key Technologies Of Machine Vision Inspection For Micro-components Under The Weak Light

In order to realize the high-precision and real-time inspection of static and dynamic status for micro-components during the process of micro-manipulation and promote the development of non-contacted micro machine vision inspection technology, combining with National High-Tech R&D Program "research on key technologies of three-dimension remote manipulation for micro-components based on ultrasonic radiation force", the measurement of the static and dynamic parameters of micro-components under the weak light environment during the process of micro-manipulation was studied in this dissertation by means of the theory and technology including machine vision, digital image processing and wavelet transform et al. On the basis of researching on micro-vision image processing and micro-vision inspection system calibrating techniques, the wavelet-based sub-pixel edge detection method and the minimum bounding rectangle measuring method were used to achieve the accurate measurement of the shape and size, orientation angle and centroid location of micro-components in 2D image. And combining with the phase correlation and optical flow estimation techniques, a new high-precision technique based on micro-vision was developed to detect multi-scale micro-displacement. Meanwhile, a set of dual-channel micro-vision inspection system was designed to meet the needs of micromanipulation and microassembly applications. The detail work was presented as follows:Chapter 1, a survey of the importance about research on micro machine vision inspection technology was given. The review of current status and future development trend of micro machine vision inspection technology was summarized. Based on discussing and analyzing the advantages and disadvantages of the traditional micro machine vision techniques, the research content of this dissertation was presented and each chapter of this dissertation was arranged.Chapter 2, according to the analysis of micro-vision images features under the weak light, the SCLAHE image enhancement method with the global and local characteristics and a kind of real-time segmentation method for micro-vision images with uneven illumination based on multi-resolution threshold were proposed. Then compared with the traditional image processing methods, the presented methods were proved to be more accurate and robust.Chapter 3, the research on visual imaging theory and aberration of micro machine vision system was carried out. According to the established micro machine vision calibration model, a kind of calibration method for micro machine vision system was proposed, which was realized by means of polynomial distortion compensation and direct linear transform (DLT). Then the experiment was demonstrated to show the feasibility and validity of the presented calibration technique.Chapter 4, a kind of 2D size measuring method for the micro-component using wavelet-based sub-pixel edge detection technique and a kind of angle and centroid measuring method for the micro-components based on minimum bounding rectangle were developed. Then combining with the phase correlation and optical flow estimation techniques, a new high-precision technique based on micro-vision was proposed to detect multi-scale micro-displacement. Finally, the computer simulation and experiments were studied to show that the proposed method was able to coordinate effectively the contradiction between the measuring range and accuracy.Chapter 5, according to the micro machine vision inspection theory presented in above chapters, a set of dual-channel micro-vision inspection system for detecting the static and dynamic status of particles was developed. Then the overall program of this system was provided, and the hardware and software of this system were designed. After experiments, the feasibility and validity of the developed techniques and system were showed.Chapter 6, the main work and conclusions in this dissertation were summarized and the prospect for further research objects was put forward.