Line search corner detection method

Abstract

The invention relates to a line search corner detection method which mainly comprises the steps that: a function for judging corner priority and a secondary filter are created; and the corner priority is applied to the estimation of the minimum distance of corners. The algorithm is an improvement to an existing line search corner detection method, can control the number of corners while maintaining the corners which are easily identified, so that the corners are uniformly distributed on an image, the follow-up image matching, image splicing and other operation are benefited, and a better effect is achieved in actual application.

Description

technical field [0001] The present invention relates to the field of computer vision, and more specifically, to a line search type corner detection method. Background technique [0002] As a key technology in image fusion, moving target detection and tracking, optical flow calculation and three-dimensional reconstruction, corner detection technology can effectively reduce the amount of information data while retaining important features of graphics and images, and can greatly improve image processing. speed, and thus widely used. At present, the methods of corner detection can be mainly divided into two types based on image edges and based on image grayscale. The effect of the former has a greater dependence on the edge extraction algorithm, and the calculation is complex and time-consuming. The latter can be divided into two categories, one is based on image derivatives such as the Plessey algorithm, the advantage is simple and easy to implement, the disadvantage is that ...

AI Technical Summary

Problems solved by technology

The effect of the former has a greater dependence on the edge extraction algorithm, and the calculation is complex and time-consuming

The latter can be divided into two categories, one is based on image derivatives such as the Plessey algorithm, the advantage is simple and easy to implement, the disadvantage is that the calculation is complicated and the positioning accuracy is not ideal; the other is based on image gray contrast relationship such as line search corner detection method, the method is simple, takes less time, and its comprehensive performance exceeds some classical algorithms, but in practical applications, the algorithm needs to give a suitable threshold t to obtain an ideal corner point, because if t is too small, it will cause Too many corner points are detected, and clustering is easy to occur, which will affect the later analysis and processing speed; if t is too large, the number of corner points will be small and the distribution will be uneven, which may make the later analysis and processing work unable to proceed

Images