High Realistic Method Of Generating Texture Based On Image

As the core of the virtual reality technology,highly realistic simulation is an important research subject in computer vision area and also has broad prospect in various fields.Generally,there are two parts to realistic simulation modeling: three-dimensional reconstruction and restoration of visual appearance.At present,For the acquisition of the target geometric model,generally divided into two kinds,(1)The active method such as the three-dimensional scanning instrument.With the continuous development of photoelectric distance measurement technology,using 3D scanning technology to obtain the target model of geometric three-dimensional information,because it can accurately reflect the objective law,which has become a reconstruction of the geometric model with the most effective and most efficient way.(2)The passive method,which is based on the image-based threedimensional modeling method,has been widely studied because of its low cost,flexible use and so on,as the data supplement of the first method.Through the above more mature system can accurately obtain the target three-dimensional geometric data,so under such conditions,the visual appearance of the advantages and disadvantages of recovery has become a decisive indicator in the high realistic simulation scene.The key to visual appearance recovery is the texture resources and the appearance through the texture and the details of the shape.Based on the above reasons,the use of 2D picture for surface texture generation has a good application prospects.In this paper,it studies the texture generation,optimization and use of the measured image.(1)Firstly,this paper analyzes the development of high realistic simulation and texture mapping at home and abroad,and analyzes the general method of texture mapping,then analyzes the method of obtaining target surface texture by two-dimensional image,(2)This paper studies the texture extraction and optimization based on the RGB color coding index table as the mapping method,which is based on the single image and UVW map.In this method,first it calculates the similarity of the contours based on the Hu moments,which is the contours of the photo of the target and the virtual model.With this step,the best match of the contour matching can be find.Then according to the characteristics of the color coded index table,Extract the texture data of the target in the two-dimensional measured image.(3)The multi-view to the target image can often contain more target texture information,because the texture resource sequence of the target obtained through multi-view is generally complementary,so this paper extracts the surface texture from the multi-view real shot.And then through the image fusion technology to fuse the texture information on pixellevel,which can achieve the purpose of optimizing the texture resources.(4)Finally,due to the theoretical modeling of the laser spot is not only time-consuming calculation,and also the get a low realistic,so this paper through the real shot of the laser spot to extract its texture to improve the appearance,then according to cg language parallel computing,so that it has efficiency and high real sense.