Kinect-based night vehicle flow statistic and vehicle model identification method

Abstract

The invention relates to a Kinect-based night vehicle flow statistic and vehicle model identification method. According to the method, a kinect depth image and a virtual coil algorithm are combined; the Kinect depth image is pre-processed, so that a moving target depth image and a hole depth image are obtained; a virtual coil is set; an integral image is utilized to generate corresponding one-dimensional signals in the range of the virtual coil; weighted combination is performed on the one-dimensional signals, so that the expression of vehicle motion characteristics is obtained, and counting is performed; in the range of combined counting signals, calculation is performed on the basis of the moving target depth image and the hole depth image, so that the geometric characteristics of a vehicle target are obtained; and the size model of the vehicle is effectively identified through an SVM (support vector machine). With the Kinect-based night vehicle flow statistic and vehicle model identification method of the invention adopted, vehicles can be better detected, and vehicle flow can be better put into statistics, the problem of inaccurate statistics caused by a condition that the night characteristics of color images are few in the prior art can be solved, complicated characteristic extraction and tracking can be avoided, and operation time can be reduced. The Kinect-based night vehicle flow statistic and vehicle model identification method has the advantages of low cost, easiness in maintenance, high real-time performance and the like.

Description

technical field [0001] The invention relates to a nighttime traffic flow statistics method, in particular to a Kinect-based nighttime traffic flow statistics and vehicle model identification method. Background technique [0002] The rapid development of intelligent transportation systems has played a key role in improving traffic conditions and improving urban modern management. Traffic flow detection and vehicle classification are important components of intelligent transportation systems. The real-time traffic flow and road condition information of vehicles provide reasonable guidance for the traffic department to operate and manage roads. When the traffic flow is too large, timely emergency measures can be taken to effectively alleviate traffic congestion, and can also guide people to choose roads reasonably to facilitate driving and reduce traffic congestion. Traffic congestion, thereby reducing the incidence of traffic accidents. Traffic flow information is an importa...

AI Technical Summary

Problems solved by technology

[0005] For the deficiencies in the prior art, the technical problem to be solved by the present invention is: provide a kind of Kinect-based night traffic statistics and car model identification method, this method adopts Kinect depth image and virtual coil algorithm, realizes traffic statistics and size model The classification of the two-dimensional image is obtained through Kinect. The depth data obtained has the characteristics of high stability and the ability to restore the real scene. Compared with the color image, the information obtained is more abundant, and it can better detect vehicles and statistics. Traffic flow avoids the problem of inaccurate statistics caused by the use of color images at night with fewer features in the prior art. The virtual coil algorithm is used to generate continuous counting signals using the continuity of vehicle targets, avoiding complex feature extraction and Tracking reduces computing time and has the characteristics of low cost, easy maintenance, and good real-time performance

Images