A method estimating vehicle under raininess

Abstract

At present, due to the needs of the society and the development of science and technology, surveillance camera systems are widely deployed in big cities. Ho Chi Minh City is building a surveillance camera system stretching over 3,800 lanes (3,600 km long) and 1400 intersections in the area. The surveillance camera systems are used to monitor and assess the security of traffic in public places. Severe traffic congestion caused serious damage to the economy. Ho Chi Minh City alone suffered about VND 27,000 billion in damages each year due to traffic congestion. As a result, surveillance systems are deployed primarily for the purpose of monitoring people and vehicles, so that supervisors can analyze and evaluate the situation within the surveillance scope of the camera. With the current technology, surveillance in our country is still largely based on people, but at one point a person can focus only on limited surveillance cameras. Therefore, building the system can detect the media density in the image so that it can be regulated, managed and guaranteed public security is a difficult and urgent problem. In recent years the machine has learned a lot, especially in the computer vision area. Derived from urgent needs in this field and the general trend of research in the world. There are many methods out there, but with a particular problem, there is only one method of optimization. Accordingly, based on publications on journals and journals, there is a need to develop a framework that integrates a variety of methods so that researchers can conveniently find the best solution for their own problems. fast. From there, there is a premise for deepening and optimizing the chosen method. After studying and developing a framework, the results show a positive direction for optimizing the population estimation problem. At the same time, it opened up many new development directions for the development system. Keywords: Crowd Estimation, Density Estimation, Raininess Weather, Local Features Extraction, Global Feature Extraction.