Research On 3D Reconstruction Based On Panoramic Camera Continuous Image

Panorama technology in recent years developed a new visual technologies, such as Tencent released SOSO street map services as well as the Chinese version of Google street map. known as "my show China." Because immersive 360-degree panoramic images strongly to the viewer the feeling of an immersive, therefore, on how to improve the three-dimensional computer image of the panoramic image restoration and processing power has received increasing attention industrial, scientific and other fields. In this paper, a continuous panoramic camera images to study the three-dimensional configuration mode, designed to restore the physical image through multiple consecutive scenes and three structures;By using the panoramic camera onboard Ladybug3 through city streets collecting image data from the extraction and matching pre-processing, feature points of the image, exercise recovery structure, dense reconstruction of three-dimensional image reconstruction process of building research, experiments and finally get three-dimensional point cloud dense city streets in a three-dimensional model of the building, and a clear point cloud structure reconstructed from the three-dimensional image in the sequence of the building.This paper mainly studied from the following aspects:(1) Research and panoramic camera onboard equipment installed on city streets for image acquisition, image pre-processing on the city streets, select the target image reconstructed city streets in need.(2) By analyzing Scale-invariant feature transform SIFT algorithm and improved the algorithm, realized for feature point extraction and matching between multiple images, multi-image geometry for calculating the feature points by matching pave and elaborated image reconstruction need to use the camera geometry, the geometric relationship between the dual image analysis and so on.(3) Select the appropriate three-dimensional reconstruction algorithm to analyze the algorithms for three-dimensional reconstruction between multiple images, including the movement to restore the structure of SFM (structure from motion) algorithm, CMVS (clustering multi-view stereo) algorithm. PMVS (patch-based multi wiew stereo) algorithm, and through visual SFM results show the three-dimensional structures of dense point clouds, the experimental results show that the reconstructed three-dimensional point cloud dense enough, well expressed in the reconstruction of the city’s streetscape target the real situation.In this paper, the content for the urban streetscape panoramic camera captured images of large-scale urban renewal quickly propose a new approach.