Recent Reviews on Dynamic Target Detection Based on Vision

Background: Vision-based dynamic target detection is an important research topic in computer vision, which is the basis for intelligent behavior analysis and event detection. Further research on dynamic target detection methods can help improve target detection and tracking mechanisms while also driving the development of other related fields. Hence, conducting a review on vision-based dynamic target detection is very significant. Objective: There are many methods for dynamic target detection. This paper introduces their classification, characteristics, advantages, disadvantages and development trends. Method: This paper reviews recent patents and representative articles on dynamic target detection in simple visual and complex contexts. The crucial methods of these references are introduced from the aspects of algorithm, innovation, and principle. Results: This paper analyzes and compares the existing dynamic target detection methods, summarizes their characteristics, main applications, and advantages and disadvantages in the current development stage, and discusses the future development and potential problems of dynamic target tracking methods. Conclusion: Vision-based dynamic target detection can accurately extract moving targets from the scene. Due to its inherent complexity, each detection method has its advantages and disadvantages in specific scenes. Currently, the research mainly focuses on the real-time robustness and accuracy of the algorithm, which needs to be further improved in the aspects of algorithm innovation, multialgorithm fusion, multi-target recognition, and algorithm adaptability. Therefore, relevant research patents and documents should be put forward, initiating the future development trend.