Human Detection Using Color and Depth Information by Kinect Based on the Fusion Method of Decision Template

Yang Shi, Xiucheng Dong, Decai Shi, Qiuyan Yang
2016 Innovative Computing Information and Control Express Letters, Part B: Applications  
As it is hard to fulfill the people detection tasks in complex environment using a common camera, this paper presents a Multi-Features human detection method based on color and depth image provided by Kinect. Firstly, we calculate the Histogram of Oriented Gradient (HOG) and Histogram of Oriented Depths (HOD) features of color and depth data. Then, the Support Vector Machine classifiers were trained based on the two features. Finally, Decision Template (DT) is calculated to match the decision
more » ... ofile of new incoming objects by the Squared Euclidean distance similarity measure. The method mentioned above is implemented in the weak light and complex environment. Experimental results show that the method has better robustness in these environments with the help of the depth images than only using color images. The detection effect has improved and the false detection and the miss rate were reduced.
doi:10.24507/icicelb.07.07.1567 fatcat:q4slzpi4h5ccvkwd7q2im5stcm