Research And System Realization Of Human Body Detection Method Based On HOG And IOELM

The detection of objects in images is an important research topic in the field of computer vision,and the human detection is an important content of image target detection.It is use the computer to simulate the human brain's way of thinking to find the human body region from the image or video.At present,human detection technology is widely used in the field of intelligent video surveillance,vehicle driving system and virtual reality.This paper focuses on the human detection method based on statistical point of view,main contributions of the paper are summarized up as follows:Firstly,the paper proposing a human body detection method based on histogram of oriented gradient and optimization extreme learning machine.The gradient direction histogram method is used to extract the characteristic value of the image.Then,use optimization extreme learning machine algorithm to classify the characteristic value of the extracted image.Finally,the non-maximum suppression method is used to accurately mark the target human body region.Experiments show that this method has a significant improvement in accuracy compared to the traditional oriented gradient and extreme learning machine algorithm,and it is almost 20 times faster than the classical histogram of oriented gradient and support vector machine method in the algorithm running time.Secondly,proposing an efficient active set algorithm for the optimization of optimization extreme learning machine,namely Improved optimization extreme learning machine.This method is employed to add several constraints to the active set at each iteration.Steepest descent direction can be obtained by adjusting the appropriate stepsize;then,predictive valuation method is used to reduce the phenomenon of repeatedly iterations.Experiments show that this method not only reduces the computational cost of training process of human body detection,and reduce the training time.Thirdly,design and implement the human detection system by the method mentioned above.The system mainly consists of three modules:image interface module,human body detection module and performance evaluation module.Image interface module include input and save images,human body detection module contains several classic human detection methods and performance evaluation module is the evaluation index of several methods.