The Research And Application For View Reconstruction Of Architecture Based On Image

Image-based on Modeling and Rendering techniques (IBMR) is a new way to produce realistic rendering. The core idea is to generate a new image from the known images. At present, the technology of generating new image based on IBMR and reconstructing the scenes has drawn great attention. Flexible, convenient, instant and accurate, this technology has a great prospect in application. This paper deals mainly with the application of IBMR on reconstruction of the building view. For the shape of the building has certain regularity, and this regularity, together with the special approach can help us create new views on the basis of all or partial information in the scene gained from the images.In this paper, digital photo-based regular building images are processed and new views are generated. The main contents of this paper include:First, the view reconstruction based on two images. In this paper, interpolation technique is used to form an intermediate view based on two images from different visual points. In this process, we should first detect the feature point of the image with Harris operator, and then match the corresponding points in the two images. With the RANSAC algorithm based on Simpson distance, we then get the optimal set of matching point by estimating the fundamental matrix, and then solve the plane homography. As to the different planes of the image, corresponding plane homography should be applied to get the middle view.Second, the view reconstruction based on a single image. In this paper, CSS corner detection operator is used to extract the feature points of the image. Since the information in a single image is insufficient, interactive operation is needed to get the feature points and solve the plane homography. Then the surface of the single image can be recreated. As to the regular building image, usually at least two vanishing points can be gained from it. In this thesis, by the approach based on vanishing point, we solve inner and outer parameters of the camera, to accomplish camera calibration, and thus recreate a single image.Third, the image inpainting. Image reconstruction may sometimes lead to a lack of image information. In this paper, fast inpainting algorithm is used to repair the missing parts of the images, which then restore the visual effects.