Method for improving automatic identification and tracking speed of target image

Abstract

It is a method to improve the object image automatic identification and tracing speed in the image technique process technique, which comprises the following steps: first to make the leap visit to the image with several picture element from left top coordinates; to turn to tracing computation once it realizes the identification; otherwise to random choose another point to enter the next circle; to adjust the leap length according to the front circle of matching approximations.

Description

technical field [0001] The invention relates to a target image processing method used in the technical field of image processing, in particular to a method for improving the automatic recognition and tracking speed of target images. Background technique [0002] An image signal is a one-dimensional signal in time and a two-dimensional signal in space, and its data volume is huge. If it is limited by data storage capacity or transmission bandwidth, the system needs to compress the image to reduce the data volume. At the same time, high-dimensional digital processing operations are complicated, so naturally very high requirements are placed on the efficiency and speed of the algorithm. In addition, the digital expression of image signals shows its unique diversity, and different systems choose different ways to express the color of each pixel. The system chooses a digital expression method to complete the main processing work, and at the same time realizes the conversion of s...

AI Technical Summary

Problems solved by technology

Although the literature has considered that image recognition must develop in a comprehensive direction that is invariant to rotation scale, parallel, accurate, and stable, and involves both the overall image and the local internal phase information of the image, but this technology must rely on the accuracy of continuous refining algorithms. In order to reduce the influence of the starting position of the third-order correlation and the change of the correlation vector on the change of the image state vector, it is also necessary to rely on the refinement of the algorithm to reduce the number of training samples and data storage

Therefore, this technology has not yet achieved rapid identification in the true sense, not to mention that the method of extracting training samples is difficult to achieve in many practical applications, such as real-time monitoring and military situations.