Research Of Traffic Flow Parameters Detection Algorithms Based On Video

The collection of real-time traffic flow parameters, such as traffic load, average travel speed, vehicle classification, and lane occupancy, plays a critical role in the advanced traffic management system and traveler information systems. Technological innovations have given rise to different types of traffic detectors. Conventional detectors, for example, inductive loop, detectors using ultrasonic, microwave, or infrared, have been put into use for several decades. Meanwhile, a promising approach, videl-based measurement system, has developed quickly. Since it has many advantages, for instance, wider-area detection and superior flexibility, many researches have been done in this field.Previous methods are mainly based on 2-dimensional image processing algorithms, especially on vehicle extraction and vehicle tracking. This paper focuses on developing a flexible and reliable system to detect the traffic flow parameters through image sequences. The vehicle detection is based on the roadway, which sets a virtual line and a virtual-loop area, like the inductive loop sensor, to detect the traffic flow parameters. The key idea of the system is converting the 2-dimensional digital image to 1-dimensional detection signal by virtual detection line preprocessing, which simplify computation. The trait of this method is only processing small image area within virtual line or virtual-loop and avoiding vehicle tracking in 2-dimensional image, hence the time cost of calculation and the computation burthen is reduced.Virtual-loop can be freely set in the video image with its positon and size adjustable. Each virtual-loop's output signals mainly derive from the pixel difference between current image frame and background within the virtual-loop area. In processing, improved background subtraction method saves vehicle information efficaciously. By comparing the common edge detection algorithms, the magniloquent Kirsch edge detection algorithm is selected to process the result of background subtraction. The result of edge detection is denoised by mathematical morphology method. Shadow edge restrained arithmetic and detection-line(area)-basedbackground updating method are presented. Shadow edge restrained arithmetic reduces vehicle shadow edge efficaciously. Background updating precision is improved by presented background updating method.An algorithm which is used to estimate the displacement of a moving object in spatial domain according to its Fourier transform spectrum is also proposed in the paper. Based on shift theorem of Fourier transform and auto-registration, the algorithm directly estimates the translation with the phase spectrum difference between continuous images of a moving object in polar coordinate system. Vehicle velocity v=s/At (where s is the displacement of vehicle and At is the interval between two continuous images) is calculated corresponding. It can give the displacement of a moving target with a resolution less than 1 pixel, hence it can be used to measure vehicle velocity with high precision.Using the algorithms depicted in this paper, the traffic flow video files in fine day and cloudy day are experimented and the experimental results show that the detection accuracy is rather high.