The Implementation Of 3D Face Imagding System And Research Of The Key Technology

Three-dimensional reconstruction techniques in computer vision has been greatly developed, three-dimensional reconstruction as a key step stereo matching, matching the achievement of results play an important role. In areas such as dental reconstruction model for higher accuracy requirements, the existing matching algorithms degree of precision reconstruction model has some shortcomings, need to match the appropriate strategy to achieve and improve the existing matching.This article studies the main application areas for the reconstruction of oral medicine model, application scenarios for such precision and speed reconstruction models have higher requirements, the traditional matching methods are often at a loss, thus matching policies and calculation methods proposed herein, complete match between the pair of images, the purpose is to avoid the drawbacks of conventional matching algorithm, to ensure matching speed, while improving the accuracy of the reconstruction model, to achieve the refinement of the matching results.A new algorithm improvements Slant is presented in this paper,The Slant oblique plane algorithm uses block-matching algorithm to complete the calculation of the matching process, as the difference is mainly targeted dealing with similar human face irregular smooth plane, in a small block like area in the calculation to the selected pixel as the center treats it as an inclined surface to be calculated, which are not considered to be within the same plane with the adjacent pixels, this calculation method is reasonable to change the depth of slower regional match results error is small, the reconstruction model can be close to the maximum extent the real object.(1) In this paper, the pyramid matches manner Slant oblique plane matching algorithm calculation method combining the pyramid matching multiple matches for processing image has obvious advantages, the pyramid of the completion of two matches each layer is calculated by matching algorithm Slant the process depends on the difference, and the first match calculated as the difference between the next match, as the initial disparity, the last layer, iteratively until the first layer, matching calculation is completed, the three-dimensional reconstruction of the refinement of the human face model.(2) Slant oblique plane matching algorithm when processing the image data does not like, does not require human intervention or adjust the image position, even in the picture there is a certain angle rotation and position offset, can be accurately capture and effective pixels, to achieve image matching purposes. This approach applies to most of the processed image template matching, with broad prospects for dynamic information matches also have a certain value.Improved calculation method by matching four pyramid matching, matching constraint conditions, using Slant oblique plane block matching algorithm to calculate the apparent difference in the matching process, the pyramid of the completion of two matches each layer, and the present times calculated as the difference between the next match, as the calculation of the initial disparity, iteratively, until the last one, to complete the matching process, to obtain a three-dimensional object depth map, obtained three-dimensional point cloud data model, and as a basis for a subsequent operations such as data fusion, drawn fine three-dimensional model, to restore the true 3D face data. Algorithm and its improvement measures proposed in this paper is mainly used in soft tissue facial reconstruction of medical applications, by using the same image data, in the same set of system reconstruction reconstruct a three-dimensional face model groups were analyzed by quantitative indicators.