Research On Moving Target Detection And Tracking Technology Based On Video Image

The detection and tracking of moving objects is a research hotspot in related fields such as computer vision,image processing,and pattern recognition.Its core is the effective analysis and processing of video image sequences.In recent years,the detection and tracking technology of moving targets has been rapidly developed,further breakthroughs have been made in related theories,and new methods proposed in conjunction with other disciplines have been widely used in practical engineering projects.The study of moving target detection by domestic and foreign scholars has increased the level of intelligence in video analytics.However,with the continuous expansion of video analytics applications and the continuous improvement of people's intelligence and accuracy requirements for mass data analysis,moving target detection still has many theories and technologies in the face of more and more complex practical application scenarios.The key issues need to be further resolved.In this paper,the background subtraction ViBe algorithm is deeply studied.Unlike traditional methods,ViBe uses the first frame to build the background model and randomly updates the samples.The stochastic strategy preserves the smoothing period of decay of the sample exponent.However,when using the first frame to set the background,the ViBe algorithm will recognize it as the foreground when the first frame is displayed.For a long period of time,the area of interest cannot be submerged in the context of causing "ghosting" phenomena.In addition,irregular flickering pixels in the background also interfere with the foreground detection results,which is not conducive to subsequent detection.In order to solve the above two problems,this paper proposes an improved algorithm based on ViBe.Using the real-time characteristics of the inter-frame difference,the ViBe result image and the three-frame difference image are combined to reduce the “ghost” phenomenon.Then,the adaptive threshold is used instead of the no-change threshold,which can suppress the effects of light and jitter.Therefore,the detection performance has been improved.It is understood that various tracking algorithms currently fall into roughly two categories: deterministic and random.After years of efforts,various tracking algorithms have been proposed for various tracking scenarios.Due to the complexity and variability of tracking,some existing tracking algorithms are weak in long-term effective tracking.Therefore,many innovative tracking schemes have been proposed for increasingly complex tracking scenarios.Based on previous literature,this paper uses differential evolution algorithm in the genetic algorithm to apply to target tracking.This is a prevalent genetic algorithm.In this paper,in order to better adapt to the target tracking,some improvements have been made in differential evolution algorithm.Specifically,introducing two disadvantaged individuals into the mutation phase further enriches the diversity of the population and accelerates the evolution of offspring.This paper also performed multiple images preprocessing,and constructed a target adaptive Gaussian Mixture Model to handle intricate tracking scenes.Experimental results show that the tracking algorithm based on improved differential evolution shows higher tracking accuracy and faster tracking speed in several challenging tracking scenes.