A5056724052- Image and Signal Denoising Methods
- Advanced Image Fusion Techniques
- Advanced Image Processing Techniques
- Medical Image Segmentation Techniques
- Advanced Computational Techniques and Applications
- Image Processing Techniques and Applications
- Advanced Algorithms and Applications
- Advanced Topics in Algebra
- Biometric Identification and Security
- Remote Sensing and Land Use
- Advanced Optimization Algorithms Research
- Face and Expression Recognition
- Anomaly Detection Techniques and Applications
- Advanced Operator Algebra Research
- Image Enhancement Techniques
- Handwritten Text Recognition Techniques
- Matrix Theory and Algorithms
- Image and Object Detection Techniques
- Text and Document Classification Technologies
- Face recognition and analysis
- Image Retrieval and Classification Techniques
- Photoacoustic and Ultrasonic Imaging
- Video Coding and Compression Technologies
- Image and Video Quality Assessment
- Interactive and Immersive Displays
Xianyang Normal University
2012-2024
Xidian University
2007-2023
Chang'an University
2021
University of Science and Technology Beijing
2019
Guangdong University of Technology
2007-2018
Ocean University of China
2017
Chinese Academy of Sciences
2010-2011
Xinyang Normal University
2011
Shanghai Normal University
2005-2010
Huazhong University of Science and Technology
2006
Level set based active contour method has been proposed for iris segmentation in recent years, but it can not converge to contours real applications because of its sensitivity local gradient extremes due the complex texture. In this paper, a novel scheme is remove before using level directly. Firstly, we use two orthogonal ordinal filters obtain robust map. Then localize region on map by an improved Hough transform. After that, Semantic Iris Contour Map generated combining spatial...
This paper presents a system that can recognize handwritten words expressed using broken letters of the Persian alphabet. The proposed be used for most activities related to gathering public information. Statistical features separated/broken are employed in system. Each letter is recognized interconnected fuzzy neural network. advantages this method include high precision owing strength network algorithm and possibility extending dataset instance codes simple manner. At last, an evaluation...
Three different sensors, a near-infrared spectrophotometer (NIR), machine vision system (MV), and an electronic nose (EN), were combined for non-destructive quality detection of 'Fuji' apples. The intention was to take advantage the three one performing local measurement physical property fruit (sugar content) others global assessment other properties (color, size, shape, aroma), combine those types (local global) improve accuracy assessment. EN used assess rotting stage apples based on ANN...
Finger pointing on touch screens is a very natural way of human computer interaction. However, for many capacitive sensing devices, it may suffer from the nature direct input since size fingers and lack precision make absolute positioning screen difficult, especially multi–touch devices. Even if high–resolution/precision devices become popular in market, cost high algorithm not flexible module. In this paper, we present Touchware, software–based technique to overcome these limitations with...
This paper proposes a human action recognition (HAR) algorithm based on convolutional neural network, which is used for semaphore motion recognition. First, collecting datas in three scenarios and Deep Convolution Generative Adversarial Networks (DCGAN) to implement data enhancement generate the dataset (DataSR). Then, 1*1 3*3 convolution kernels are design full network model further compressed using group obtain new HARNET. Experiments show that mAP of HARNET DataSR 94.36%, size 76M, 30%...
In this paper, we introduce and study the generalized inverse <TEX>$A^{(1,2)}_{T,S}$</TEX> with prescribed range T null space S of an adjointable operator A from one Hilbert <TEX>$C^*$</TEX>-module to another, get some analogous results known for finite matrices over complex field or associated rings, operators.
A network model for facial expression recognition is designed and named DI-FERNet in this paper. The uses depth-wise separable convolution, dilated convolution residual module to build the structure. This paper MTCNN perform face alignment processing on pictures dataset. large number of experiments are carried out selected datasets KDEF RAF. test accuracy 97.2% RAF 77.1%.
In this paper, we present a new image denoising method for removing Gaussian noise from corrupted by using shearlet transform and nonlinear diffusion.The is decomposed the to obtain coefficients in each subband; then diffusion scheme based on statistical property of used shrink noisy coefficients.The test shows proposed has better performance compared relevant methods.
Local binary pattern (LBP) is an important image texture feature extraction method. This paper proposes a combination of adaptive threshold and directional local representation, to realize the retrieval. composed two parts: 1)The standard deviation gray value all pixels in neighborhood center pixel are used as comparison for encoding. adaptively changed following grayscale changes neighborhood, can indicate severity scale change; 2)directional pattern: by comparing different directions,...
The Perona-Malik (PM) model, an effective anisotropic diffusion, can preserve edges while removing the noise. However, disadvantage of PM model is tending to impair textures and details so that denoising not sufficient in whole process. For this reason, we present a novel texture preserving (TPPM) models based on local directional variance. In TPPM diffusion coefficient adaptively determined be low large variance domain high domain. related parameters are studied. Comparative results real...