Geometric Correction for Projection Image Based on Azure Kinect Depth Data

Abstract

In order to improve the universality of the projector to the irregular projection surface environment, a geometric distortion correction algorithm for projection images based on Azure Kinect depth data was proposed. The algorithm used Azure Kinect to collect the color information and depth information of the irregular projection surface, obtained the largest visible projection area from the color image, and calculated the target projection position according to the ratio of the projection area. Then the natural feature matching set was divided into regions according to the depth information layering rules of the irregular projection surface, and the unknown regions without natural feature coverage in the depth information were clustered to the known hierarchical regions. Finally, the homography matrix was obtained by solving each depth layered plane, and the set of homography matrix was constructed to correct the geometric distortion of the projection image, so as to obtain the expected projection image without distortion. Through the comparison and analysis of the experimental data, it could be seen that the correction algorithm can not only obtain a projection display that meets the human visual perception, but also correct the feature error between the output image and the original image to the sub-pixel level.