Research On Modeling And Mapping Technology Of Details Properties On Real Surface

In the area of computer graphics, an important part is to render images with realistic effects. The lighting model is the basis for the creation of realistic images. To calculate and render the color of geometric model surface points need to specify the following parameters of those points and substitute them into the model for further calculation, such as the reflectance properties, normal, the positions of light sources and viewpoints, incident light intensity, etc. This process is known as forward rendering, which renders objects in the scenes based on the known information. In order to obtain more accurate realistic graphics, the textures details of the surface should be taken into consideration.With the development of our society, the application areas, which need to reproduce the three-dimensional shape and image of the real object, become wider than before, such as digital museums, heritage and digital tourism, etc. To describe and render the specific real object into virtual environment need to obtain and model detail properties from real surface. It is difficult to specify a large number of parameters to render the appearance of the real objects in the process of forward rendering. Lighting, reflective properties or geometric parameters can be reversed from measurement data, real scene images and other information. Such process or sampling data-driven rendering is called reverse rendering. Reverse rendering is the inverse process against the traditional rendering process, which is raletated to computer vision. As the reverse information can accurately reflect the information of the appearance parameters of real scenes, in the process of forward rendering in traditional graphics the information can be used to render highly realistic images and reproduce real object in the computer. In recent years, the research of modeling and rendering of surface texture has made significant progress. However, the modeling of complex surface texture, especially the modeling of the object that is complex in geometry and reflectance properties, is still confronted with a great challenge.The obtaining and modeling material detail properties of objects is still a problem remaining unresolved in the world as well as a difficult and hot topic.Based on this background, this dissertation conducts a research on how to get real object detail properties through inverse rendering and to apply the detail properties to the forward rendering algorithm. Based on the review of domestic and foreign research on the surface reflectance model, this dissertation summarizes the algorithms of obtaining reflectance properties and detail geometry through inverse rendering. This dissertation puts forwards the methods of the acquisition and modeling of surface detail properties through inverse rendering, and the rendering algorithms through texture mapping on the basis of the physical knowledge of surface reflectance and the basic theory of computer graphics. This dissertation focuses on the study of the acquisition, modeling and rendering of details geometry and reflective properties, which aims to establish an integrate, applicable and effective method to solve the problems. The specific contents of this research run as follows:1. Expressions of the details geometric and the reflection properties.The study of this part is carried out through the division of surface detail properties of real objects into geometry features and reflectance properties. Surface reflectance properties are divided into energy reflection properties and the wave reflection properties whose expressions are studied respectively in this dissertation. The basic lighting model BRDF (Bidirectional Reflectance Distribution Function) is decomposed to the n-polynomial based on the incident light vector component (x, y) with the polynomial coefficients as the energy reflection properties. The wave reflection properties suggest that the surface of the object absorbs and reflects the light of different wavelengths, that is, the color character of the surface. To reduce the influence on the geometric shadow for sampling color on the surface, we have carried on a research on the theory and methods of the uniform color map obtained under uniform light environment to express wave reflection property, which provides the basis for the uniform color map to substitute for light source in order to calculate reflection energy and to specify color for surface. The expressions of details geometric are also studied. Using the "the synthetic normal" to approximately describe the direction of small and complex structure patches covered by sampling pixels.2. The methods to acquire and store the data of surface details geometry and reflection properties for real objects.This part focuses on the acquisition methods of synthetic normal from a patch on the surface and uniform light color of the object, and the method to fit with the coefficients of quadratic polynomial. Furthermore, the sampling programs are framed and details properties sampling devices are designed according to the characteristics of diffuse reflection and specular reflection. To facilitate the details properties data using texture mapping techniques in forward rendering, we've studied the ways to store the digital images of polynomial coefficients, synthetic normal and uniform light color.3. The application of the details properties data to forward rendering.The study of this part mainly focuses on the texture mapping method of uniform light color map, polynomial coefficients map, synthetic normal map and especially the attempt to find a solution to coordinate system transform for polynomial coefficients map and synthetic normal map. We also study algorithms of forward rendering with details properties that can control the rendering result in the calculation of light.As the objects vary widely in nature, it is a big challenge to model the detail properties especially for the object with complex geometry and reflection properties. This dissertation is not intended to seek a general solution to all circumstances, but focus on some certain similar issues.The main innovation in this paper includes the following:1. This paper proposed a polynomial expression of the reflection BRDF model based on the incident light vector component (x, y). Polynomial coefficients maps describe reflection energy characteristics. The parameters of the basic lighting model is simplified in this study. When viewpoint is fixed and according to geometric of the surface has invariability, the reflection BRDF model can be decomposed into a binary n-polynomial of incident light vector x, y components in the way of Taylor decomposition. Through the experments we proposes that light reflectance BRDE model should be approximated by the binary quadratic polynomial of incident light vector x, y components and the errors of rendering results are within our vision. The sample BTF photos are captured when viewpoint is fixed and the location of light source changes. The quadratic polynomial equations system is established to solve quadratic polynomial coefficients by means of least square. The coefficients can be stored to images which are called polynomial coefficients maps which can describe energy reflection properties.2. This paper puts forward that uniform light color maps can describe the wave reflection properties of the object's surface, that is, the object's color properties.In this paper, the concept of uniform light color map is put forward to describe wave reflection properties of the object's surface. Uniform light color map is captured under the uniform light conditions of surface light source, which records the characteristics of reflecting light-wave (color) for the surface points, that is to say, the proportion among different wavelengths of light (energy ratio of RGB of sample) remains unchanged. Meanwhile, the relative differences of reflecting capability are recorded with different luminances of different points under the same light conditions. In the process of forward render, the uniform color map can substitute light source to calculate reflection luminances and specify color for surface.3. In this study, the concept of synthetic normal map is proposed to describe details geometry property.In this paper, we put forward the concept of synthetic normal to describe details geometric property of the surface, based on the fact that a pixel is a sample for a small patch of surface with small and complex structure. Two algorithms are raised to evaluate synthetic normal:the photometric stereo measurement method and the BRDF diffuse quadratic extremum method.4. A series of combination algorithms for light rendering based on the uniform light color map, coefficient map and synthetic normal map are put forward in this study. The texture mapping algorithms for details properties maps, and the algorithm of self-defined coordinate system that can transform light vector into texture space are also proposed. The uniform light color map, polynomial coefficients and the synthetic normal map are digital images that are easy to adopt the technique of texture mapping. The pixels data among the maps are corresponding to each other. Therefore, the mapping can be completed at one time by using this texture mapping. Because the data of polynomial coefficients and synthetic normals are related to sampling coordinate system, we define an intermediate coordinate system which can transform light vector into texture space. And this makes it possible for the lighting calculation of each pixel in texture space rather than in the model space, which can reduce mapping errors and enhance the speed. Finally, several blending rendering algorithms with details properties maps are put forward too, which can adjust the results of the rendering freely.In short, based on the theory of optical physics modeling, computer vision and computer graphics, this dissertation puts forward the arithmetics and technical framework of obtaining, modeling and mapping details properties, which are appropriate to the known light conditions. It may provide new methods and technologies for the modeling and rendering of real surface details. The results of this research will have theoretical significance and practical value in the fields of 3D movies, games and virtual reality, etc.