Visual detection method of multi-movement target positions in large view field

Abstract

The invention discloses a visual detection method of multi-movement target positions in a large view field, which comprises the steps of design of a color mark plate, splicing of a double-camera view field, searching and positioning of a mobile target, conversion of an image coordinate system and a world coordinate system, and the like. The color mark plate is used for identifying the mobile target and determining target position data and designed to include an excircle-type color mark blocks with different colors and equal radius; the view field splicing method comprises the steps of shooting by using two cameras and splicing two scene images overlapped with each other in real time to acquire the large view field; the searching and the positioning of the mobile target comprises the steps of searching the mobile target based on a combined strategy of global grid searching and flight path searching and positioning the mobile target by means of seed point area growth; and finally the position and the posture of the mobile target in the world coordinate system are acquired by using a conversion model based on the image coordinate system and the world coordinate system which are relatively measured. The invention has the advantages of simple method, small calculated quantity, higher positioning accuracy and low application cost.

Description

Technical field [0001] The invention relates to a method for visually detecting the position and attitude of a moving object, in particular the invention relates to a method for detecting the position and attitude of multiple moving objects in a large field of view. Background technique [0002] The visual detection technology of moving targets organically combines image processing and information science to form a technology that can identify targets and extract target information from image signals in real time. It is a relatively new dynamic target pose detection technology. The main purpose of moving target visual detection is to determine the precise position and attitude of the target in the field of view, that is, to locate the target. [0003] The current common positioning methods can be summarized into the following categories: inertial positioning, landmark positioning, global visual positioning, etc. Inertial positioning methods often include cumulative errors a...

AI Technical Summary

Problems solved by technology

[0004] In the case of a large moving target working space, the use of a single camera to achieve target pose detection usually has a large image distortion. Even if some distortion correction methods can be used to correct it, this will increase the computational complexity, and the effect may not be good. Reach the established detection accuracy

Images