Unlicensed vehicle detection method based on AdaBoost and SVM (support vector machine)

Abstract

The invention relates to an unlicensed vehicle detection method based on AdaBoost and an SVM. The method comprises the following steps: 1), an original video sequence image is subjected to high-expansion downsampling in the rank direction on a DSP (digital signal processor) platform, so that an RGB (red, green and blue) three-channel color image is obtained; 2), aiming at RGB three-channel color image, the marginal information and color information of the RGB three-channel color image are comprehensively considered so as to synthesize a gray level image; 3), an AdaBoost detector is adopted to detect the unlicensed vehicle in a video sequence aiming at the synthesized gray level image ; and 4), a nonlinear secondary SVM divider is established for detected targets and background, and the false alarm probability is further reduced. The unlicensed vehicle detection method based on the AdaBoost and the SVM can quickly judge non-target samples in a large quantity of candidate detection windows and are applied to detection of targets in the video sequence successfully, and the detection time is shortened by about one third.

Description

technical field [0001] The invention relates to an intelligent traffic detection technology, in particular to an unlicensed vehicle detection method based on AdaBoost and SVM. Background technique [0002] Intelligent Transportation System (ITS) started from the computerization of traffic management in the 1960s and 1970s. , Comprehensive transportation management system that plays a role in all directions. [0003] Vehicle detection in video sequences is an application of moving object detection in the field of intelligent transportation. Currently commonly used vehicle detection algorithms based on adjacent frame image information are: [0004] (1) Inter-frame difference method: This algorithm is to subtract the gray value of the corresponding pixel of the two frames of images before and after. If the gray value difference is small, it can be considered that there is no car passing by the point; otherwise, the gray value changes greatly, then Think a car is passing by. ...

AI Technical Summary

Problems solved by technology

[0007] Due to the efficiency of the algorithm itself and the limitation of the DSP frame rate, the above three algorithms have certain limitations in practical applications. For example, the inter-frame difference method and the optical flow method are not suitable for occasions with low frame rates. The false detection rate of the time-background subtraction method increases significantly

[0008] In addition, the current vehicle detection algorithm based on single-frame image information relies too much on license plate information and template matching between license plates, and it is basically impossible to detect unlicensed vehicles

Images