Visual target rapid detection, tracking, parallel computing and fusion method based on multi-kernel DSP

Abstract

The invention discloses a visual target rapid detection, tracking, parallel computing and fusion method based on a multi-kernel DSP. The method comprises the steps of for a complex environment of a visual target in a high speed motion process, training an efficient offline classifier for rapid detection and locking of a target; realizing high speed target tracking through adoption of a kernelizedcorrelation filtering tracking algorithm after the target is locked; effectively fusing a Kalman filtering operator, a detection operator and a tracking operator through Kalman filtering predictionand confidence degree comparison policy judgment, thereby forming a multi-kernel tracking algorithm; taking a high master frequency multi-kernel DSP applicable to image data parallel computing as an embedded program transplantation carrier, and allocating different operators to multiple kernels of the DSP according to a task; and accelerating image processing speed through adoption of a multi-kernel parallel computing technology and a software pipeline loop. The method is applicable to visual target rapid detection and tracking airborne and vehicular embedded processing equipment, and patricianly applicable to occasions on which target motion speed is rapid, computing quantity is high, and a demand for timeliness and accuracy is relatively high, under complex scenes.

Description

technical field

[0001] The invention relates to an embedded visual tracking technology, in particular to a multi-core DSP-based fast detection and tracking parallel computing and fusion method for visual targets. Background technique

[0002] Embedded visual tracking is the application of digital image vision technology in the embedded field, including machine vision, computer vision, video signal processing, pattern recognition, embedded applications and other disciplines, intelligent tracking or positioning equipment, medical diagnostic equipment, Industrial automation equipment, etc. are increasingly dependent on this technology. In recent years, new algorithms have been emerging in visual object tracking. Accurate tracking often means a large amount of calculation. However, due to the contradiction between the calculation amount of the algorithm and the processing speed of the hardware platform, the two Key Performance Indicators: The balance between real-time and accur...

Images