A Specific Algorithm Based on Motion Direction Prediction

Zhesen Chu,Min Li,Wei Wang

doi:10.1155/2021/6678596

Abstract

In this paper, we study the estimation of motion direction prediction for fast motion and propose a threshold-based human target detection algorithm using motion vectors and other data as human target feature information. The motion vectors are partitioned into regions by normalization to form a motion vector field, which is then preprocessed, and then the human body target is detected through its motion vector region block-temporal correlation to detect the human body motion target. The experimental results show that the algorithm is effective in detecting human motion targets in videos with the camera relatively stationary. The algorithm predicts the human body position in the reference frame of the current frame in the video by forward mapping the motion vector of the current frame, then uses the motion vector direction angle histogram as a matching feature, and combines it with a region matching strategy to track the human body target in the predicted region, thus realizing the human body target tracking effect. The algorithm is experimentally proven to effectively track human motion targets in videos with relatively static backgrounds. To address the problem of sample diversity and lack of quantity in a multitarget tracking environment, a generative model based on the conditional variational self-encoder conditional generation of adversarial networks is proposed, and the performance of the generative model is verified using pedestrian reidentification and other datasets, and the experimental results show that the method can take advantage of the advantages of both models to improve the quality of the generated results.

Highlights

In recent years, human behavior understanding, as a key task in intelligent applications such as automated driving, service robots, and advanced surveillance systems, is one of the hotspots of interest for researchers in the field of computer vision [1]
We study human motion targets contained in HEVC video coded streams, based on information such as motion vectors partially decoded as a database. e HEVC compressed encoded video processing with the human motion target as the object provides an in-depth study of detection and tracking techniques
Since the GMM hybrid Gaussian model algorithm is susceptible to interference from light intensity, there is a lot of noise when detecting human motion targets in the images, and the detection rate of human targets is relatively poor due to the large difference between the human body and the background. e VIBE algorithm is relatively poor at detecting the integrity of the human motion target

Summary

Introduction

Human behavior understanding, as a key task in intelligent applications such as automated driving, service robots, and advanced surveillance systems, is one of the hotspots of interest for researchers in the field of computer vision [1]. Predicting the trajectory of pedestrians is of great practical significance; for example, in the safety-critical task of automated driving, only by correctly deducing the intentions of pedestrians around the vehicle and accurately predicting their future trajectories can the vehicle plan its path, avoid obstacles, and prevent crashes from occurring [7]. Due to the complexity of human behavior and the diversity of the surrounding environment, accurately predicting the trajectory of pedestrians is a challenging task [8]. Due to the complexity of human behavior and the diversity of the surrounding environment, accurately predicting the trajectory of pedestrians is a challenging task [8]. e changes of pedestrian trajectories are determined by a variety of factors such as the goal intention, the direction and location of the surrounding

Methods

Results

Conclusion