Highlight Removel Combined Separating Reflection Components And Pixel Filling

With the rapid development of modern industry, the structured light 3D measurement is widely used in the field of computer vision, reverse engineering,product development, quality control, and the accuracy requirements are continuously improving at the same time. But when measure specular objects by the structure light, the information collected will miss because of the highlight, and it will lead to the failure of the measurement. So how to solve the three-dimensional measurement of specular objects has become the focus of research in recent years.On the basis of the previous theories, this paper improves the traditional method of highlight removel, and applies it in the field of structured light 3D measurement. The main research contents are as follows:1. We analysis of the effect of specular reflection on structural light three dimensional measurement, and then design the general planning of the system based on the theory. The overall scheme includes the selection of the highlight removel method, the selection of the system calibration method, the construction of the experimental platform and so on.2. On the basis of the existing highlight removel, a new highlight suppression method is proposed which combine reflection component separation and pixel filling.We study the theory of the reflection component separation, and the method can make the highlight region to form a cavity, and it can not achieve the purpose of removing the highlight. Therefore, this paper makes a study on the theory of pixel fill,and combines the reflection component separation with the theory of pixel filling.3. Three dimensional reconstruction experiments of different objects are carried out by the method of highlight suppression proposed in this paper, and then the number of pixels in the image before and after processing are statistically compared.From the contrast and experimental data we can see, the highlight area in the image disappeare, the number of specular pixels generally reduces by 30-40 times andsurface reconstruction and the measured surface can reflect the measured surface shape accurately. Both theoretical analysis and experimental results show that the proposed method is feasible and effective.