Method for identifying, tracking and converting target based on confidence degree and multi-frame judgement

Abstract

The invention discloses an integrated method for identifying and tracking a target based on confidence degree and multi-frame judgement, comprising the following steps of: (1) inputting an image; (2) carrying out histogram drawing on a grey-scale image; (3) identifying the position of the target in the image, and giving the confidence degree of the identification result; (4) determining whether the position of the identified target is consistent or not, and classifying the identification results with the same position into a class; (5) calculating the correct probability that a certain position belongs to the target according to the confidence degrees of the identification results and continuous multi-frame identification results by utilizing the Binomial distribution theory; and (6) selecting one image with the highest correct probability as an initial reference frame in a tracking stage and taking the identification result of the image as the reference position of the target. In the invention, the confidence degree of target identification results and the identification results of continuous multiple frames are subjected to comprehensive judgement, thus the operating mode of an automatic target identification system can be automatically, accurately and reliably converted into the target tracking stage from a target identifying stage.

Description

technical field [0001] The invention belongs to the field of automatic target recognition, and specifically relates to a method for target recognition tracking conversion, which uses the confidence degree of the target recognition result and the recognition results of continuous multi-frames to make a comprehensive judgment, so that the working mode of the automatic target recognition system can be automatically determined by the target recognition phases are accurately and reliably converted to target tracking phases. Background technique [0002] To complete the automatic recognition of the target, the computer vision system often needs to go through the stages of detection, recognition and tracking. Among them, the detection and recognition stage realizes the task of discovering and confirming the target, and the algorithm in this stage often takes a lot of time. At the same time, due to the influence of image quality, pose or scale changes, it is impossible to guarantee...

AI Technical Summary

Problems solved by technology

In the conversion process, recognition errors may occur. In this case, how to avoid the impact of the wrong recognition results on the tracking stage, and select the correct recognition results ...

Images