Method and device for tracking image target based on parallel particle filtering

Abstract

The invention discloses a method and a device for tracking image target based on parallel particle filtering. The method comprises the following steps: (1) acquiring an image sequence and defining F as the NO. F frame of the image sequence; (2) acquiring a to-be-tracked target; (3) distributing particle points and ensuring that F is equal to 2; (4) inputting the NO. F frame; (5) carrying out brightness filtering till obtaining a column diagram; (6) comparing the particle points with the target column diagram to acquire normalized weight; (7) acquiring the coordinates of the target in the NO. F frame; (8) outputting the NO. F frame and updating the target column diagram; (9) resampling the NO. F frame; (10) judging whether to continue tracking and entering step (11) if continuing tracking, otherwise, stopping tracking; and (11) ensuring that F is equal to F plus 1 and returning to the step (4) to repeat the step (4) to the step (10). The method and the device break the logical dependence of sampling, weight calculation and resampling in the prior particle filtering algorithm, and realize algorithm parallel processing and execution acceleration by utilizing the normalized weight of the previous frame as a resampling function.

Description

technical field [0001] The invention relates to the technical field of machine vision, in particular to an image target tracking method and device based on parallel particle filtering. Background technique [0002] Machine vision is the use of machines instead of human eyes for measurement and judgment. The machine vision system refers to converting the captured target into an image signal through the image capture device, and transmitting it to a dedicated image processing system, and converting it into a digital signal according to the pixel distribution, brightness, color and other information; the image system performs various processing on these signals. Operation to extract the features of the target, and then control the on-site equipment action according to the result of the discrimination. [0003] In machine vision, the tracking of a specific target is a very important part of it. In recent years, tracking systems based on particle filters have become a research ...

AI Technical Summary

Problems solved by technology

First of all, the three steps of the particle filter algorithm have logical dependencies, and the existing methods need to execute these three steps in sequence; secondly, although the distributed particle filter method realizes parallel processing, there is still a need for processor-based operations. The problem of trivial access to storage space and sequential execution; finally, there are places for further exploitation of parallelism at various steps in the processing process

[0008] Generally speaking, the existing particle filter-based tracking system uses a distributed processor array to meet the real-time performance of the algorithm, but there is still a lot of room for parallelization and acceleration.

Images