Duanquan Xu

A New Algorithm for Shoreline Extraction from Satellite Imagery with Non-Separable Wavelet and Level Set Method

 Abstract—An effective and precise method for shoreline detection from satellite imagery is presented. The algorithm is based on two main steps: (1)the detection of singularities in a single image using non-separable wavelet and (2)amendment procedure using distance regularized level set evolution scheme. Firstly, by selecting appropriate parameters, the non-separable wavelet filter banks which can provide information of different orientations are used to capture the singularities of the selected single satellite image; Secondly, obtaining the modulus image by utilizing sub-images decomposed from the non-separable wavelet filter banks; Thirdly, extracting the shoreline iteratively with the use of distance regularized level set evolution scheme. Experiments are conducted and results show that the proposed algorithm is applicable to satellite imagery, and the shoreline is robust to noises as well as blurring.

Fingerprint enhancement based on non-separable wavelet

Fingerprint enhancement is necessary for low quality images. In this paper, we develop a new method for fingerprint enhancement based on non-separable filter banks. We first decompose the fingerprint image using the non-separable filter banks, which can decompose the fingerprint image efficiently and obtain wavelet coefficients, then modifies the coefficients by applying the adaptive approach to reduce the noises and increase the contrast between ridges and valleys according to the geometry feature of images. Then we apply the inverse wavelet transform to map the result and use a two-dimensional median filter to reduce noise and preserve edges. Experiments have been conducted on the fingerprint database FVC2002 in our study. The results demonstrate that the proposed method works well in fingerprint enhancement without under-enhancement and over-enhancement, and it is more effective and robust than other existing methods.

Study on the Quantitative Cytometry and Cervical Cancer Diagnosis Technology Based on Support Vector Machine

The cervical cancer screening technology based on quantitative cytometry is studied. Feulgen stain is conducted on the sample of cervical tissues. Then the microscopic image of the sample is captured by CCD camera. The images of cell nucleuses are extracted by image segmentation. And the morphological, optical density and texture parameters of the cell nucleuses are calculated. The dimension of the feature parameter vectors is reduced using F-score and Random Forest algorithms. And the types of the cell nucleuses are identified by a SVM classifier. The diagnosis whether the carcinogenesis exists or not is given according to the distribution of DNA content of the epitheliums.

Dollar bill denomination recognition algorithm based on local texture feature

In this paper, a dollar bill denomination recognition algorithm based on local texture feature is proposed. this paper proposes a local texture feature dollar denomination recognition algorithm, this algorithm first use the between-cluster variance method about the dollars local image binarization to enhance the effect of differences, and then through the cross algorithm to extract the local texture feature, which makes the recognition work correctly. The simulation results show that the method is fast, high precision, suitable for real-time face recognition.

Global sparse partial least squares.

The partial least squares (PLS) is designed for prediction problems when the number of predictors is larger than the number of training samples. PLS is based on latent components that are linear combinations of all of the original predictors, it automatically employs all predictors regardless of their relevance. This will degrade its performance and make it difficult to interpret the result. In this paper, global sparse PLS (GSPLS) is proposed to allow common variable selection in each deflation process as well as dimension reduction. We introduce the ℓ2, 1 norm to direction matrix and develop an algorithm for GSPLS via employing the Bregmen Iteration algorithm, illustrate the performance of proposed method with an analysis to red wine dataset. Numerical studies demonstrate the superiority of proposed GSPLS compared with standard PLS and other existing methods for variable selection and prediction in most of the cases.

Learning a Sparse Representation for Robust Face Recognition

Based on the assumption that occlusions have sparse representation on the nature pixel coordinate, Sparse Representation based Classification (SRC) [9] adopts an identity matrix as occlusion dictionary to deal with the occlusions or noises. However, this assumption is often violated in real applications, such as the faces are occluded by scarf. In this paper, we present an approach to learn an occlusion dictionary from the data. Thus, the occlusions have sparse representation on the learned occlusion dictionary and can be effectively separated from the occluded face images. Experimental results show our approach achieves better performance than SRC, while the computational cost is much lower.

Efficient Group Learning with Hypergraph Partition in Multi-task Learning

Recently, wide concern has been aroused in multi-task learning (MTL) area, which assumes that affinitive tasks should own similar parameter representation so that joint learning is both appropriate and reciprocal. Researchers also find that imposing similar parameter representation constraint on dissimilar tasks may be harmful to MTL. However, it’s difficult to determine which tasks are similar. Z Kang et al [1] proposed to simultaneously learn the groups and parameters to address this problem. But the method is inefficient and cannot scale to large data. In this paper, using the property of the parameter matrix, we describe the group learning process as permuting the parameter matrix into a block diagonal matrix, which can be modeled as a hypergraph partition problem. The optimization algorithm scales well to large data. Extensive experiments demonstrate that our method is advantageous over existing MTL methods in terms of accuracy and efficiency.