Subjects: Computer Science >> Integration Theory of Computer Science submitted time 2019-01-03 Cooperative journals: 《计算机应用研究》
Abstract： This paper developed a hybrid convolution neural network for perceptual crowd counting, which could accurately predict density maps in extremely crowded scenes. It consists of merely two components: the front-end is a dilated convolutional neural network to extract two-dimensional features; the back-end deployed a fractionally strided convolution to lower the loss of image information caused by down-sampling. This paper designed the model structure based on the dataset Shanghai Tech, then in an attempt to acknowledge and analyze the performance of the algorithm, , and afterwards made use of the evaluation indicators of the regression problem, the average absolute error (MAE) and the mean-square error (MSE) as the criteria. Additionally, testing the method on Shanghai Tech (MAE=100.8) , UCF_CC_50 (MAE=305.3) and WorldExpo'10 datasets while the experiment results reveal that the proposed model can effectively reduce MAE and MSE when compared with previous methods.
Subjects: Computer Science >> Integration Theory of Computer Science submitted time 2018-05-18 Cooperative journals: 《计算机应用研究》
Abstract： In order to identify the copied images accurately and quickly, this paper presented an image hash algorithm based on CS-LBP(centrally symmetric local binary pattern) texture and bit image statistics. Firstly, it preprocessed the image, and obtained the approximate image and high frequency information by using the three-level wavelet decomposition. It segmented the second and third approximations image by Ring and extracted the statistical characteristics of each ring. The horizontal component and the vertical component of the second and third level high-frequency information execute bit image decomposition, which extracted statistical features. Finally all the low-frequency and high-frequency features together to generate image hash sequence. The experimental results show that the classification performance of this algorithm is better than some existing hash algorithms and has good accuracy in copy detection applications.
Hits
Hits
Downloads
Comment