Extraction And Visualization Of Boundary Surfaces Within 3D Images

Boundary surfaces of interested objects are very important characters within 3D images. According to these characters, objects can be measured, analyzed, interactively operated, recognized and traced. Therefore, extraction and visualization of 3D images boundary surfaces is one of most important problems in 3D images analysis. Boundary surfaces can be classified continuous triangle surfaces understood by consecutive implicit surface contained in its sections based the definition of 3D regular data and discrete boundary surfaces composed of voxels (cubes). As a result, this dissertation discusses extraction and visualization of continuous triangle surfaces and discrete boundary surfaces.Firstly, with interactive operation, special objects of interest can be extracted by a tracing algorithm based on connectivity of edge surfaces. In our framework, firstly, high gradient ROI (Region of Interest) and low gradient ROI on objects of interest need to be chosen to calculate high and low gradients thresholds. And then based on cells including implicit continuous surfaces (called as edge cubes) as seeds, we perform our tracing algorithm to extract all edge cubes containing complete boundary surfaces. Finally, triangles of boundary surfaces can be extracted from edge cubes by Marching Cubes and be visualized. Our algorithm combines interactive operation and tracing algorithm to extract special object of interest and solve the problem that gradient thresholds are difficult to be set. Moreover, efficient data structure is designed to save time when tracing algorithm is performed.Secondly, a novel extraction algorithm of discrete boundary surface is proposed based on the fact that discrete edge voxels lie on both sides of implicit continuous surfaces, we present a new method to detect and trace discrete edge surfaces. Firstly, cells including implicit continuous surfaces are detected and traced. Then, in these cells, every side which intersects implicit continuous surfaces is found. And finally, discrete edge surfaces are composed of vertexes of all these sides. Our proposed method can avoid the difficulty of direct trace of discrete edge voxels and extracted accurately and completely edge surfaces with transitional gradient.Finally, to better visualize boundary surfaces with gradient values changing from high to low within 3D images, semi-transparent visualization is proposed. The algorithm exploits intelligence and knowledge of human beings to distinguish boundary surfaces with low gradient values from cracks resulted from noises by visualizing them translucently. Therefore, boundary surfaces with gradient values changing from high to low within 3D images can be visualized completely and understood correctly by users.The proposed interactive triangle surface extraction of special object of interest, tracing algorithm of discrete boundary surface and semi-transparent visualization can extracted complete and accuracy boundary surfaces with 3D images. These methods have already applied in a great number of biological, medical and industrial 3D images, and the experimental results show the effectiveness and advantages of these two algorithms.