Dense three-dimensional scene reconstruction method and system based on existing map

The invention provides a dense three-dimensional scene reconstruction method and system based on an existing map, and relates to the technical field of computer vision, and the method comprises the steps: S1, carrying out the feature extraction of the existing map; s2, performing feature extraction on frames of the common camera according to a rough pose provided by the common camera in a scene scanning process; s3, calculating the accurate pose of the common camera by using the features in the existing map according to the features in the common camera frame; and S4, adjusting and optimizing the obtained accurate pose, and reconstructing a dense three-dimensional scene through the optimized pose. According to the method, the accurate pose of the common camera can be calculated by using the existing accurate map information, and adjustment and optimization are carried out, so that the problems occurring during three-dimensional reconstruction of the common camera are solved, and a better reconstruction effect