Research On The People Counting Methods Based On Multi-feature Regression

In recent years, in order to make up for the shortcomings of the traditional CCTV, and change the traditional model of the post investigation, researchers have begun to use computer vision monitoring technology to automatically monitor the crowd. Crowd analysis is one of the most challenging tasks in intelligent visual surveillance system. The utility model can be used for the automatic detection of the population density level, the number of people, the abnormal detection and alarm. As one of the most important analysis methods of video frames, people counting has been widely concerned in the field of security control, and plays an important role in crowd monitoring and management.Because of the existence of many flexible factors, such as irregular, nonrigid, occlusion and so on, people counting has been a difficult problem in the field of intelligent monitoring.On the basis of consulting a large number of relevant literatures,this paper makes a study on the method of people counting based on multi-feature regression.A people counting algorithm based on foreground blob and KAZE feature is proposed. To get started, the foreground blob is obtained by using Gauss background modeling and morphological filtering techniques to the GAMMA corrected images. Next, the features of foreground blob, such as weighted area, perimeter length and perimeter orientation histogram are extracted. Then each blob image and the original image are smoothed by Gauss, and the intersection of these two smoothed images is done. After the mask and image enhancement operation, the more clear and accurate edge contour feature can be obtained. At the same time,the KAZE algorithm is used to obtain the key point features of each blob image, in order to get more relevant feature points. Finally, these features are used as input into SVR regression model.The experimental results show that the proposed algorithm has a high accuracy of crowd counting.A people counting algorithm based on improved Gauss process regression is proposed.First of all, according to the above algorithm, the features are extracted. Next, the square exponential covariance function is selected as kernel function. The bacterial foraging algorithm is used to optimize the hyper-parameters to obtain the optimal solution ,and then the regression model is established. The experimental results show that the proposed algorithm which makes use of bacterial foraging to optimize the hyper-parameters can obtain better parameters and improve the accuracy of the people counting.