Target matching detection method

Abstract

The invention discloses a target matching detection method and belongs to the technical field of intelligent video monitoring. The target matching detection method combines a particle swarm optimization algorithm with classifier training to obtain feature selection (feature dimension reduction) parameters and an optimal classification model in one time, and accordingly the training time consumption is lowered; the feature selection parameters are used to perform the image feature dimension reduction at a test stage, and the optimal classification model is used for classifying to obtain a final target matching detection result. According to the target matching detection method, the space distribution information of color is added to color features, and the space distribution information is fused with an improved histogram of oriented gradient (HOG) feature to generate an initial image feature. Compared with the prior art, the target matching detection method is capable of effectively improving the target matching detection precision and greatly reducing the processing time.

Description

technical field [0001] The invention relates to a target matching detection method, which belongs to the technical field of intelligent video monitoring. Background technique [0002] Object recognition techniques based on supervised classification learning have attracted attention in recent years for object matching detection in intelligent video surveillance. The general method for target recognition is: (1) select and extract the appropriate features of the target; (2) select the appropriate classification and discrimination model based on these features, that is, establish a suitable target classifier; (3) select the target sample library Carry out the training of the classifier to obtain the classification model; (4) use the classification model to classify the samples of the test set to realize the recognition of the target. When used for post-event analysis, several samples of the target can be extracted from a video in surveillance for training, so as to identify ot...

AI Technical Summary

Problems solved by technology

[0004] The technical problem to be solved by the present invention is to overcome the shortcomings of large amount of calculation and poor real-time performance in the prior art, and provide a target matching detection method, which uses the method of combining particle swarm optimization algorithm and classifier training to realize image feature Effective dimensionality reduction and optimal classification model selection, this method can not only effectively improve the accuracy of target matching detection, but also greatly reduce the processing time

Images