摘要

In this paper, a novel depth extraction method based on the regional feature point is proposed to enhance the estimation accuracy of the depth and is used to reconstruct 3D scene. Specifically, each elemental image is segmented into many regions firstly and the depth of the object point corresponding to each feature point in each region is extracted, then the depth corresponding to all feature points can be calculated to express the depth corresponding to all object points, which can enhance the accuracy by solving the wrong matching problem in weak texture regions and reduce the amount of calculation. To segment each elemental image properly, we first preprocess each elemental image and segment each elemental image into many regions, then merge some similar regions. Some computational reconstruction experiments are carried to verify the effectiveness of the proposed method.