Approximation Calculation Of Boundary Surface Of The Interest Target In 3D Images

With the rapid development of 3D imaging technology, a lot of 3D biomedical images have been generated. In the 3D biomedical image processing and analysis, boundary surface of the interested objects is a very important feature. By extracting the boundary surface of the target object, we can take advantage of it in the object recognition, quantitative analysis and some other's. Therefore, detecting and extracting the boundary surface of the interested object is a very important issue. In this article, we will discuss how to accurately extract the boundary surface of interest.Firstly, in some 3D images, the iso-surface corresponding to a given iso-value might contain several different connected components, but only one represents the boundary surface of a structure of interest. In such cases, instead of all components of the iso-surface, we only need to extract a specific connected component from 3D image. Existing iso-surface extraction methods either cannot be used to extract a specific iso-surface component from a 3D image, or cannot correctly separate two iso-surface components that simultaneously pass through some cubes of a 3D image. Therefore, this paper proposes a new method to extract the interested iso-surface component from a 3D image. With the method, we can trace all cubes passed through by the selected iso-surface component from 3D images, and generate a polygonal surface model for the component by constructing suitable triangles in each traced cube. It can completely remove the isolated small fragments which around the structure of the interest.Secondly, when the grey values of the interest object relatively large changes, a fixed iso-surface is not suitable for use in approximating the boundary surface of the interest object. Therefore, in this paper, it propose an adaptive approximation algorithm to approximate the boundaries of the interest structures within 3D images. First of all, we divide the volumetric images into many sub-images, and then calculate the average of the gray values of discrete sampling points which on the boundary surface of the interest within the sub-image, the last to approximate a small patch boundary surface with a small patch iso-surface within sub-images. In this method, the boundary surface of the interest could be locally and adaptively approximated with different surface patches in different sub-images. Experimental results show that the algorithm is feasible and effective, and can considerably improve the approximation accuracy.Thirdly, it is important to extract the boundary surface of the blood vessels structure within MRA images. If the interest structures of the blood vessels within MRA image can be goodly approximation by iso-surface surface and can be separated from the surrounding tissue, so we extract of a connected component of the iso-surface of the interest blood vessels. If the interest structures of the blood vessels within MRA image can be goodly approximation by iso-surface surface but can't be separated from the surrounding tissue, so we extract the interest blood vessels by boundary-iso-surface algorithm. If the interest structures of the blood vessels within MRA image can't be goodly approximation by iso-surface surface, so we adaptive iso-surface approximation of the boundary surface of the interest blood vessels .The methods mentioned above have been applied to a lot of 3D biomedical images, and the experimental results reflect the effectiveness and reliability of these algorithms.