Data correlation problem processing method in large-view field photoelectric imaging multi-weak and small target tracking

Abstract

The invention belongs to the technical field of image analysis, and discloses a data correlation problem processing method in large-view field photoelectric imaging multi-weak and small target tracking. The processing method comprises the following steps that deviation between the direction of each echo point trace in a tracking door and the direction of a prediction point at the previous moment is calculated, a certain direction weight is given to corresponding echoes according to the size of the deviation, and the direction interconnection probability under the direction information is obtained; then a gray level difference between each echo point trace in the tracking door and a target point at the previous moment is calculated, a gray scale change factor of each echo is obtained, and the grey level interconnection probability under the gray level information is obtain through calculation according to the factor; and finally, different weights are given to the interconnection probability which only contains the distance information calculated through an original GPDA algorithm and the calculated direction interconnection matrix and the gray level interconnection matrix, and thetotal interconnection probability is obtained through weighted summation. The processing method has the advantage that on the basis of inheriting the original GPDA algorithm, the tracking performanceis greatly improved at a small calculation amount cost.

Description

technical field [0001] The invention belongs to the technical field of image analysis, and in particular relates to a method for processing data association problems in large-field photoelectric imaging multiple weak and small target tracking. Background technique [0002] At present, the existing technologies commonly used in the industry are as follows: [0003] At present, the most widely used data association algorithms are based on Bayesian theory, evidence theory, and intelligence (fuzzy algorithm, neural network, genetic algorithm, ant colony algorithm, etc.) theory. The association algorithm based on Bayesian theory is still the mainstream algorithm to solve the data association problem. Among them, Joint Probabilistic Data Association (JPDA) is recognized as the optimal algorithm for solving multi-objective data association problems. Therefore, in practical engineering applications, based on JPDA and combined with the particularity of the problem, some improved al...

AI Technical Summary

Problems solved by technology

[0005] (1) There are few target features in the large-field photoelectric imaging system, and there is no geometric shape for reference, which is a typical pixel-level small target; at the same time, because the target signal is weak, it is easy to be submerged or confused with noise. It is easy to generate a large number of suspected targets

[0006] (2) Although the conventional GPDA algorithm solves the calculation of the calculation of the target correlation degree in the process of multi-target data association in the sparse target scene, but in the face of dense clutter environment, there are many false targets and noise, and the calculation load is still large

[0007] (3) The conventionally discussed GPDA methods all use d

Images