Research On Virtual View Drawing Based On Depth Mapping In Panorama

In recent years,with the development of hardware technology and the development of computer vision theory,in order to enhance the user's immersion,interaction and experience,panoramic roaming technology is widely used in people's production and life.Panorama roaming technology refers to the user in the path of any location can see where the location of the panoramic image.The roaming technique consists of multiple views,and the images between the two viewpoints can be captured either by camera-intensive shooting or by drawing a virtual viewpoint,but the amount of data generated by the camera's direct acquisition method is very large,so it has not been widely used.In the drawing of virtual viewpoints,the general methods are model-based rendering and image-based rendering.Among them,the model-based rendering technology on the hardware requirements are high and with the complexity of the scene increased the complexity of the model also increased.Therefore,the research point of this paper is the virtual view drawing technology based on depth map in panoramic roaming.Therefore,this paper studies virtual view drawing technology based on depth map of image rendering in panoramic roaming.This paper focuses on the virtual view drawing technology based on the depth image and introduces proposed depth map acquisition technology and the key issues of virtual view drawing technology.Because the traditional depth map acquisition method is based on the triangular geometric similar relationship in the camera system model but in the panoramic roaming this triangular relationship is broken,that is,the original relationship in the existing model is not established.So in order to solve the problem,this paper models the camera placement system in the panoramic roaming application,and re-finds the corresponding geometric relation in the model to obtain the required depth map.Finally,the experimental results show that the depth map obtained by the method is superior to the depth map generated by the contrast method,and then the accuracy of the proposed algorithm is verified.The high quality depth map is very important for the drawing of the virtual viewpoint,and the more accurate the depth map,the more accurate the virtual view image is drawn.In addition to focusing on the acquisition technology of depth,this article also focuses on the solution of the key problems in virtual view drawing with the help of depth information,the main problem is hole and overlapping what result in image blur and distortion.In this paper,we propose a new method that determine hole pixel at first and then the hole pixel reflected to the reference image.The next step is to find the desired texture,if there is no available texture,we fill the hole by collar pixels in the virtual view image with the proposed algorithm.The virtual view image is processed by this method,a higher quality drawn image can be obtained.Finally,it shows that the proposed method can solve the problem of hole and overlap,and the virtual view image is similar to the real scene.The problem of cavitation and overlap is of great significance,it can reduce the degree of image blur,reduce the distortion of the image and improve the image quality.