The vulnerability of Deep Neural Networks (DNNs) to adversarial attacks has become an important research area of machine learning. It has been known that many state-of-the-art DNNs suffer the risk of universal adversa...
详细信息
In this paper we construct a novel human body model using convolution surface with articulated kinematic skeleton. The human body's pose and shape in a monocular image can be estimated from convolution curve throu...
详细信息
In face recognition, the dimensionality of raw data is very high, dimension reduction (Feature Extraction) should be applied before classification. There exist several feature extraction methods, commonly used are Pri...
详细信息
The particle filter offers a general numerical tool to approximate the posterior density function for the state in nonlinear and non-Gaussian filtering problems. While the particle filter is fairly easy to implement a...
详细信息
Blurred images are caused by many factors such as defocus, motion, and atmospheric turbulence. Due to the unknown various factors that cannot be distinguished in the blurred image, it is necessary to propose a unified...
详细信息
This paper presents a texture segmentation approach which is based on the Markov random field model (MRF) and feed forward neural *** texture is modeled by the second order Gauss MRF model, and the least square error ...
详细信息
This paper presents a texture segmentation approach which is based on the Markov random field model (MRF) and feed forward neural *** texture is modeled by the second order Gauss MRF model, and the least square error estimation is employed for the solution of model parameters. To perform texture segmentation, we introduced an improved BP algorithm to get faster learning speed. Experiment shows that better segmentation results can be obtained than the traditional Euclidean distance method.
This paper introduced a novel high performance algorithm and VLSI architectures for achieving bit plane coding (BPC) in word level sequential and parallel mode. The proposed BPC algorithm adopts the techniques of co...
详细信息
This paper introduced a novel high performance algorithm and VLSI architectures for achieving bit plane coding (BPC) in word level sequential and parallel mode. The proposed BPC algorithm adopts the techniques of coding pass prediction and parallel & pipeline to reduce the number of accessing memory and to increase the ability of concurrently processing of the system, where all the coefficient bits of a code block could be coded by only one scan. A new parallel bit plane architecture (PA) was proposed to achieve word-level sequential coding. Moreover, an efficient high-speed architecture (HA) was presented to achieve multi-word parallel coding. Compared to the state of the art, the proposed PA could reduce the hardware cost more efficiently, though the throughput retains one coefficient coded per clock. While the proposed HA could perform coding for 4 coefficients belonging to a stripe column at one intra-clock cycle, so that coding for an NxN code-block could be completed in approximate N2/4 intra-clock cycles. Theoretical analysis and experimental results demonstrate that the proposed designs have high throughput rate with good performance in terms of speedup to cost, which can be good alternatives for low power applications.
This paper addresses the application of hand gesture recognition in monocular image sequences using Active Appearance Model (AAM). For this work, the proposed algorithm is conposed of constructing AAMs and fitting the...
详细信息
This paper addresses the application of hand gesture recognition in monocular image sequences using Active Appearance Model (AAM). For this work, the proposed algorithm is conposed of constructing AAMs and fitting the models to the interest region. In training stage, according to the manual labeled feature points, the relative AAM is constructed and the corresponding average feature is obtained. In recognition stage, the interesting hand gesture region is firstly segmented by skin and movement ***, the models are fitted to the image that includes the hand gesture, and the relative features are ***, the classification is done by comparing the extracted features and average features. 30 different gestures of Chinese sign language are applied for testing the effectiveness of the method. The Experimental results are given indicating good performance of the algorithm.
A classifier-based method to select and fuse grey level co-occurrence matrix (GLCM), Gaussian Markov random field (GMRF) and discrete wavelet transform (DWT) features to improve texture discrimination is presented. Fe...
详细信息
Irregular pyramids are made of a stack of successively reduced graphs embedded in the plane. Each vertex of a reduced graph corresponds to a connected set of vertices in the level below. One connected set of vertices ...
详细信息
暂无评论