Part Recognition And Feature Parameter Extraction Based On CCD Scanning Information

With the development of technology and human beings'increasing needs, traditional machine vision can not meet the needs on recognizing the 3-D object. In most circumstance, it needs to adopt measuring methods and 3-D geometry modeling to transfer the object into 3-D model in computer, realizing recognition and orientation of the 3-D object.Image understanding is a major step towards Artificial Intelligence, which involves systems that will make certain decisions or perform some activities based on their interpretation of an input image. It has widespread applications in automated detection, robot self-motion snatching, automated assembly, autonomous navigation and medical diagnoses.With the application of laser scanning and triangulation method, 3-D information of object is calculated in this dissertation. Denoise is done according to continuity of graph contour. Research the methods of image segmentation. Categorize measuring data according to visibility in range direction. Different image segmentation methods are presented for different categories. Theory research has been made around model of curve fitting. Realize profile curve fitting with the application of least square method and extract the shape feature and geometry feature. Analyze the 3-D object recognition arithmetic based on shape feature and geometry feature, in which it relate to shape language description, area and inflexion information. Interpret the 3-D object according to the recognition result. Acquire the geometry information and position information of the 3-D object to realize the orientation.