Technique Study On 3D Reconstruction And Appearance Reality Based On Kinect

In recent years,with the rapid development of computer vision and virtual reality technology,3D reconstruction and realistic technology are widely used in 3D printing and film and television production motion analysis of objects and medical treatment and other fields,its research has important practical and practical value.At present,the research of 3D reconstruction,laser scanning and 3D reconstruction of binocular vision are of high cost,large amount of data and computation complexity and low accuracy problem based on Kinect has the advantages of easy operation and low price,can capture real-time depth and color information of the object surface etc..However,there are some problems in the depth image acquired by Kinect.The depth image has high noise and lack some of depth.In addition,there are errors in the process of 3D model reconstruction,such as point cloud registration and fusion,which leads to the loss of reconstructed model details.In realistic rendering using texture mapping process due to difference between the color tone and the lighting conditions,resulting in a joint effect on the surface of the model.In this paper,we first propose a method for automatic classification of depth image holes in depth image data filling algorithm.In this algorithm,the hole types in depth images are divided into smooth,and non smooth holes.For the two different types of holes,the hole filling algorithm based on SSIM(structural,similarity)weights is adopted respectively,and first order Taylor approximation algorithm based on bilateral filtering for depth image restoration,effectively filling the depth missing region,and the smooth edge of the hole in the process of image restoration is solved.On the basis of this,two strategies for enhancing the effectiveness of 3D reconstruc-tion are explored in view of the missing details in 3D reconstruction model.The adaptive weighted registration algorithm based on CSHOT(Color Signature Histograms of Orien-tations)descriptor and the TSDF(Truncated signed distance function)point cloud fusion algorithm based on depth noise model.First select the matching points effectively,calcu-late the feature points on the CSHOT descriptor,constructing adaptive weighted energy equation;The depth noise model is used to reduce the influence of noise on the fusion of point clouds,and the problem of missing detail in the reconstructed model is solved.Finally,in order to obtain the realistic appearance of 3D reconstruction model,the color fusion cube is used to define the compound weights and unify the different lines.The difference of hue and brightness between the pictures resolves the seam phenomenon in the process of texture mapping,and makes the fused 3D model have higher sense of re-ality.Based on Kinect 3D reconstruction and appearance of realistic technology research,made full use of the depth data obtained by Kinect 3D reconstruction,and combining the texture image to achieve realistic 3D rendering.Compared with the traditional methods of reconstruction and reality,the application of 3D reconstruction in practical production and life is promoted,and the research work in the field of reality is also promoted.