Motorcycle type identifying method in complex scene

Abstract

The invention discloses a motorcycle type identifying method in a complex scene, which comprises the following steps of: initializing a component dictionary of a video image, learning a parameter of each component in the component dictionary, calculating an optimal composition structure according to the learned parameters of the components and an XOR searching tree, training and integrating vehicle templates by adopting the optimal composition structure, and detecting and identifying a motorcycle type in the video image by using the vehicle templates. According to the motorcycle type identifying method, the optimal composition structure of the vehicle templates is learned by adopting a dynamic planning algorithm, the XOR searching tree and a large quantity of actual samples, thus the efficiency of training the templates is increased, better discrimination is achieved, and actual application is facilitated. According to the motorcycle type identifying method, by combining a Latent SVM (Support Vector Machine) algorithm and a robust HOG (Histograms of Oriented Gradients) characteristic, the motorcycle type in the complex scene can be processed, and instantaneity and generality are ensured.

Description

Technical field [0001] The invention relates to the fields of image pattern recognition, intelligent video monitoring and intelligent transportation, and in particular to a vehicle type recognition method in a complex scene. Background technique [0002] Vehicle recognition based on video images refers to the automatic recognition of different types of cars from images and videos, such as vans, cars, trucks, buses, etc. It is a key technology in intelligent transportation systems, no matter in the field of intelligent transportation monitoring , Still in the field of automatic toll collection of highways and parking lots, it has extremely important applications. [0003] Vehicle recognition based on video images is generally divided into three parts: 1. Segmentation of vehicle images; 2. Feature extraction; 3. Recognition and classification of vehicle types. At present, the relevant vehicle identification methods in the literature mainly include: (a) prototype-based vehicle identi...

AI Technical Summary

Problems solved by technology

Although such methods are more robust than prototype-based methods, they also have common shortcomings: first, such methods are still very dependent on the quality of image segmentation, and often only deal with the background Simple situation; second, the features selected by this type of method are not robust enough; third, the models used by this type of method are relatively simple and can only represent rough information of the target, and generally can only be divided into large, medium and small Three types, and cannot be further classified in detail; Fourth, this type of method is still highly dependent on the placement of the camera

However, since this method uses a heuristic method to initialize the parts, it cannot always find a good initialization position for the parts, which often leads to the failure of the model; secondly, the heuristic method is also highly dependent on the artificially set parts number and shape of

In practice, the number of parts and the shape of the target are often not fixed, it depends on the camera angle of view, the distance from the target and the difference between the target categories, so that for each type of vehicle, it is difficult to manually select a set of correct parts

Images