Method, system and medium for three-dimensional reconstruction of human body

Abstract

The invention belongs to the method for three-dimensional reconstruction of a human body in the field of computer technology, comprising step 1: interactively photographing parts; step 2: constructing an initial human body shape, that is, a rough model; step 3: establishing a first point cloud coordinate matrix; Step 4: Extract feature point coordinates of close-up parts; Step 5: Establish the second point cloud coordinate matrix; Step 6: Smooth fitting repair, transfer to a file in 3D model file format; Step 7: File according to 3D model file format , to complete the 3D reconstruction. The present invention can quickly construct a high-precision three-dimensional human body model of the target human body through a group of continuous shootings of different viewing distances, a collection of ordinary photos partially covering the shooting surface, and supplementing local close-up photos.

Description

technical field [0001] The invention belongs to the field of computer technology, and in particular relates to a method for three-dimensional reconstruction of a high-precision human body. Background technique [0002] The 3D reconstruction of the human body has important applications in many fields, but due to business differences, the accuracy requirements of different parts will be different, and the corresponding local fineness will also have different demands. However, the reconstruction speed is also very important for the user experience. How to accelerate the reconstruction? For parts with demanding dimensional accuracy and variable due to individual differences, it ensures efficient and fast modeling of the overall calculation, and flexibly improves the fineness of any body part according to the needs of business and individual signs. [0003] At present, there are mainly three methods on the market, (1) Scanning once with a high-precision laser, and then reconstruc...

AI Technical Summary

Problems solved by technology

These three have their own advantages, but in terms of human body size mapping, the acceptable scanning time within half a minute, and the best shooting distance of 1 meter, the unique optical components of the first type are expensive and stable in accuracy Regular manual intervention is required. After fixed installation, the single-scanning speed is the fastest, but it is susceptible to interference from external ambient light and object reflections. It needs to calibrate equipment and levels from time to time, and the global accuracy is uniformly fixed. Some weak precision requirements still provide high Accurate point cloud, thus affecting the efficiency of subsequent calculations; the second type of 3D information extraction takes a long time, cloud computing puts too much pressure on the cost of computing power unit, and cannot solve the bottleneck of computing tasks running on operation and maintenance, and the accuracy of the surface is difficult to improve, which affects information The security and continuity of transmission to the cloud have high requirements; the third type relies on proprietary infrared characteristics. Within the reliable line-of-sight range of existing infrared sensors, the accuracy of depth information of photographed objects is limited by hardware. The color dependence on the target is large, and equipment and levels need to be calibrated from time to time

Images