Study On Reconstruction Disparity Surface Based On Microscopic Stereovision And Laser Scanning

Microscopic stereovision system based on stereo light microscope has been used in microscopic measuring fields. The traditional measuring methods output the disparity surface based on the procedures of stereo matching applied to the whole or local images. However, these methods require that stereo images should contain enough and remarkable features. Meanwhile, the matching results usually contain many mismatching and error matching points. We propose a novel method for disparity surface reconstruction of small objects, which is combined with stereo light microscope and laser measuring techniques. The surface of small objects is scanned by laser fringe. And stereo image sequence containing laser stripes is obtained. We extract the central contours of laser stripes by Hessian matrix. The principles of registering the central contours are derived. A disparity coordinate system is defined, in which we analyze the relationship between the moving direction and reference plane of floor. Then the method of aligning disparity contours is presented. We use a small object with plane surface to evaluate the precision of our method. We reconstruct the disparity surfaces of four kinds of small objects with different sizes and shapes by our method and traditional block matching method. Furthermore, we compare their real shape with their measured shape, and the performance of our method with that of traditional block matching method. And the latter experimental results show that the real and measured shapes described by disparity surface are consistent based on our method, and that the performance of our method is obviously better than that of traditional block matching method. The disparity surface output by our method will be the input data for microscopic vision model, and used to accurately measure surface profile of small objects.Chapter 1 analyzes the microscopic stereovision research status and describes the contents and significance of this paper.Chapter 2 analyzes the function of microscale objective measurement. The microscopic stereovision laser scanning system is designed. Then, the appropriate hardware configuration is selected and the measurement process of systme is stipulated.Chapter 3 analyzes the characteristics of the laser stripe microscopic image. Firstly, the images are preprocessed, including segmenting image area, decomposing image and filtering image. Then, the center contour of laser stripe is extracted though subpixel method of detecting fring center.Chapter 4 analyzes the grid coordiate of left and right sample images. The special longitudinal distortion model of microscopic image is summarized. This chapter proposes the method of correcting distortion model. Finally, the effectiveness of method is verfied by much experiment.Chapter 5 define the disparity surface and space of targe objective. The arrangment rule of microscopic disparity stripes is derived from relationship between objective moving direction and the reference plane. The disparity surface of different kinds of objective are reconstructed thought experiments.