Research On The Static Reconstruction Method Of Three-dimensional Human Body Based On Java 3D

Starting from the research of static three-dimensional reconstruction, this paper proposes two different kinds of human body surface data acquisition methods. On basis of that, two kinds of three-dimensional human body reconstruction methods were proposed. One is based on Poser and the other one is based on the Visualization Human Body Data Set.The information of human body surface vertexes and the modeling of three-dimensional body are inseparable, because surface data acquisition is the foundation of human body modeling. Based on the application of Virtual Dressing, a modeling method witch combined both the advantages of Poser and Java 3D was proposed. Simply put, we use 3DS files witch were exported by Poser to gain human surface vertexes data in data acquisition stage, and when in the reconstruction stage, we make good use of Java 3D's powerful three-dimensional reconstruction and interactive control functions. By analyzing 3DS file's structure, human body surface vertexes data can be obtained. Then these data were used under Java 3D to form body surface witch was approached by small triangle surfaces. Reconstruction speed is fast and the model is realistic. For latter step realizes the three dimensional human body's parameterization dynamic deformation and the three dimensional human body's movement simulation builds the foundation.The slice images of Visualization Human Body Data Set have a widespread application in medical science. This paper uses this data set as the source of human head reconstruction research. According to the accuracy level of clothing application, we put forward our own Matlab image processing steps: image processing, image binarization, image denoising, and image edge detection. After these four steps, we can get obtained the complete continual forehead slice outline vertexes. Considered that the element coordinate extracted by Matlab can't be used directly by the float coordinate system of Java 3D, this paper presents a contour vertex data normalization method, and the re-calculation method of vertex index value. Validity and correctness has been verified in experiments. In the contour line formative stage, this paper presents the data selection algorithm based on string value. After experimental confirmation screening the outline data point connection has formed the complete contour line. In the three-dimensional reconstruction stage, this paper analyzed the characteristic of both the cam counter line space reconstruction and the non-cam counter line reconstruction. In the foundation of three heuristic algorithms, we proposed an improved shortest diagonal line algorithm.