A tracking method based on spatio-temporal tracking for extracting the center of light stripes

Abstract

The invention relates to a method for extracting and tracking the center of a light bar based on space-time tracking, comprising the following steps: using a time series model to predict the coordinates of the center of the light bar at the next moment; taking the predicted center of the light bar as a starting point, and the tangential direction of the center of the light bar is determined by the Hessian Determined by the eigenvector of the matrix, use the Hessian matrix to obtain the unit tangent vector of the point, and use the tangent direction of the light strip as the extension direction of the valley ridge area to realize the valley ridge tracking of the light strip in space. The invention can improve extraction speed and extraction accuracy.

Description

technical field [0001] The invention relates to the technical field of light bar center extraction and tracking, in particular to a light bar center extraction and tracking method based on space-time tracking. Background technique [0002] The research on feature extraction and tracking can be traced back to the late 1970s. It plays a vital role in the fields of machine vision and pattern recognition, and its extraction results directly affect the subsequent detection and recognition results. Line structured light system calibration and feature extraction and tracking are ubiquitous in the process of 3D measurement. The essence of line structured light 3D measurement is to extract the coded corner points of the rotary table to determine the rotation angle, and to extract the center of the structured light strip to locate the optical center. Reconstruct the 3D contour point cloud of the rotary parts in the acquired image sequence. How to quickly and accurately extract tracki...

AI Technical Summary

Problems solved by technology

[0004] Commonly used light bar center extraction algorithms include gray-scale center of gravity method, direction template method, curve fitting method and Hessian matrix method. The performance of each light bar center extraction algorithm is shown in Table 1. The gray-scale center of gravity method and direction template method are simple and easy. It is fast and effective, but the accuracy of the extracted light bar center is not high, and it is difficult to apply to high-precision measurement occasions; the curve fitting method has sub-pixel level accuracy, but the algorithm is complex and time-consuming, and it is difficult to meet the needs of real-time measurement; the Hessian matrix method The algorithm complexity is high, but it has sub-pixel-level precision, and can distinguish the direction of structured light, and has the best comprehensive robust performance

Images