Multi-feature-fusion-based traffic video data collection processing method

Abstract

The invention discloses a multi-feature-fusion-based traffic video data collection processing method. The method comprises the following steps: (1), carrying out virtual region division on a traffic video image; (2), calculating a foreground pixel ratio of a virtual region, determining a standard difference SDe of edge strength, and determining a vehicle detection direction angle VD; (3), calculating a vehicle existence confidence grade number according to the obtained foreground pixel ratio, the standard difference SDe of edge strength, and the vehicle detection direction angle VD based on the image processing at the step (2); and (4), carrying out CL data statistics to determine the number of monitored vehicles. According to the technical scheme, the method has the following advantages: (1), a vehicle is detected by using features like the foreground area, texture change, and pixel motion in a virtual coil comprehensively and the vehicle detection algorithm is free from the influences of the adverse weather and illumination based on effective fusion of three kinds of features; and (2), the vehicle speed is estimated according to the pixel motion vector in the single virtual coil, thereby solving an inherent vehicle wrong matching problem of the dual-coil speed measurement method.

Description

technical field [0001] The invention belongs to the technical field of road traffic video image data collection and processing, specifically a method for collecting and processing traffic video data based on multi-feature fusion, thereby overcoming the influence of complex and diverse traffic scenes and weather conditions on image data processing, and improving Accurate detection of road vehicles and vehicle speeds. Background technique [0002] In the application of intelligent transportation system, traffic management, traffic simulation and traffic flow theory research, a very important task is the collection of traffic information. At present, the information that most traffic information collection equipment can provide is mainly macroscopic parameters such as the flow rate, speed, and density of a single vehicle type. These traffic information do not distinguish between vehicle types. Collecting multi-vehicle traffic information is of great significance for traffic ma...

AI Technical Summary

Problems solved by technology

[0004] Existing commercial traffic video detectors mainly use the virtual area method, but according to its practical evaluation report, there are generally the following defects: under conditions such as car light reflection, moving shadows, bad weather, camera shaking, etc., the error rate of traffic detection data results Higher, for example, the detection error rate at night due to the reflection of car lights reaches 74.3%; the detection error rate due to rainy and snowy weather is as high as 16.3% even in the daytime, and the detection error rate at night is as high as 50%

Images