Cao Bui-Thu

Image retrieval using Contourlet based interest points

In this work we present the method for image retrieval based on the Non-Subsampled Contourlet Transform (NSCT) and the Harris corner detector. The NTSC-based interest point detector is proposed by combination of NTSC and Harris corner detector called the Contourlet Harris detector. We also present the method how to extract the image features using this Contourlet Harris detector that is applied for image retrieval. Experiments are implemented on the WANG database aiming to compare retrieval effectiveness of proposed method to some methods have announced. Results demonstrate that the proposed method shows a quite improvement in the retrieval effectiveness.

A new descriptor for image retrieval using Contourlet Co-Occurrence

In this paper, a new descriptor for the feature extraction of images in the image database is presented. The new descriptor called the Contourlet Co-Occurrence is based on the combination of the contourlet transform and the Grey Level Co-occurrence Matrix (GLCM). In order to evaluate the proposed descriptor, we perform the comparative analysis of existing methods such as Contourlet [2], GLCM [14] descriptors with the Contourlet Co-Occurrence descriptor for image retrieval. Experimental results demonstrate that the proposed method shows a slightly improvement in the retrieval effectiveness.

Texture image retrieval using Phase-based features in the complex wavelet domain

The phase holds crucial information about image structures and features, but only the real part or the magnitude of the transform coefficients is often used for image processing applications. In this paper, a method for the feature extraction of images called Phase-based LBP is presented. Proposed method is based on the combination of phase of complex wavelet coefficients and the Local Binary Pattern operator (LBP). We also perform the comparative analysis about retrieval effectiveness of phase information of some complex wavelet transforms using for Phase-based LBP. Experimental results, achieved with the standard rotated Brodatz dataset, show the interest of this method comparing with another methods only based on the real part or the magnitude of the wavelet coefficients for texture image retrieval.

An efficiently phase-shift frequency domain method for super-resolution image processing

How to reconstruct a high resolution and quality from low resolution images captured from a digital camera is always the top target of any image processing system. Exploit the aliasing feature of sampled images, we propose a new technique to register exactly the motions between images, including rotations and shifts, by using only frequency domain phase-shift method. Based on registered parameters, the precise alignment of input images are done to create a high-resolution by using interpolation methods. It is more exactly when we compare our algorithm to other algorithms in simulation and practical experiments. The visual results of super-resolution images reconstruced by our algorithm are better than that of other algorithms. It is possible to apply our algorithm to increase resolution of digital camera or video systems.

An improvement of curvelet based super-resolution image processing implemented on ARM AT91SAM9RL

This paper copes with the problem of improving the quality of the curvelet interpolation in super-resolution image reconstruction. The curvelet interpolation has been proposed by some authors, however the quality of reconstructed images from their implementation is not as high as expected and the processing time is also not efficient. To improve the curvelet interpolation, a 2-stage interpolation algorithm in the curvelet domain combined to a filtering step for the reconstruction is proposed. The interpolated images are compared with images provided by other previous High-Resolution reconstruction methods and to the ideal interpolation. The experiments in Matlab and hardware based approach show the appropriate improvements of PSNR and MSE in comparison with the other previous methods.

Curvelet transform based Super-Resolution image processing

Super-Resolution (SR) Processing is a technique that produces a High - Resolution (HR) image from a range of Low - Resolution (LR) images. There are many methods for implementing the SR signal processing. In this paper, we introduce a specific transformation that were developed in the recent years called Curvelet Transform (CT), and then apply it in SR Processing to obtain a quality improvement of SR images. We discuss two separate algorithms using the Discrete Curvelet Transform (DCT) and then compare them based on the results of HR images. We find out that the first algorithm, named the iterative algorithm can produce better HR images than the second one, the Interpolation algorithm. However, the iterative algorithm also take more computational cost than that of the Interpolation Algorithm. It is a reasonable trade-off between images quality and processing speed. We also made a comparison between our algorithms and previous works such as the Projection onto Convex Sets (POCS) and the Nearest Neighbourhood algorithms to show the quality improvement.

Scene-based video super-resolution with minimum mean square error estimation

Motion estimation is a key problem in video super-resolution (SR). If the estimation is highly accurate then the high resolution (HR) frames reconstructed is better in quality. Otherwise with small errors in estimation, they will create more degradation in the reconstructed HR frames. In many recent studies, the motion estimation is applied on every block of pixels. There is too little input data for estimating process so that it is hard to get high accuracy in results. This paper presents a new method for SR video image reconstruction through two main ideas. First, video frames are separated into two sections, as scene and motive objects. The motions of the scene are the same and uniform. We will have much data for estimating, so that the result can be more accurate. Second, the motion estimation is based on three parameters, rotation and shifts in vertical and horizontal. It presents a perfectly estimating for real motion of camera when capturing video frames. Based on that, an efficient algorithm is proposed by combination of block matching search method and minimum mean square error estimation. The results of the proposed algorithm are more accurate than those of other recent algorithms. It can be easy to see one we visualize the HR video frames reconstructed by other algorithms.

An efficient approach based on Bayesian MAP for video super-resolution

Multi-frame super-resolution brings out much potential to reconstruct real high-resolution video sequences. This potential is achieved based on its capacity to combine missing information from different input low-resolution frames. Although there have been many studies in recent decades, super-resolution problems for real-world video processing still have many challenges. This is dues to two problems of: how to address the affecting factors: motion, sampling and noise explicitly and how to solve them exactly and efficiently. This paper introduces an efficient approach for video super-resolution by addressing real motion, sampling and noise models. Based on that, we proposed a model for receiving a practical video and an efficient framework to estimate adaptively the motion and noise to reconstruct the original high-resolution frames. Our system achieves promising results when compare with other state-of-the-art in quality and processing time.

LTE indoor MIMO performances field measurements

Long-term evolution (LTE) and multiple input multiple output (MIMO) have earned reputations to be a cutting-edge technology, which can boost significantly wireless communication performances. The paper aims at providing LTE MIMO performances in indoor environments and, therefore, guidelines for network operators can be proposed. Medium access control throughput (MAC TP) and some system parameters in LTE network that are linked with MAC TP, such as Channel Quality Indicator (CQI), Modulation and Coding Scheme (MCS), Ranking Indicator (RI), Pre-coding Matrix Indicator (PMI), as well as MIMO utilization, are analysed. Effects of indoor propagation, Line of Sight (LoS), No-line of Sight (NLoS), strong and weak signal levels on Signal to Noise Radio (SNR) strength and MIMO utilization are clarified. In this paper, the performances of MIMO transmission mode over transmit diversity (TxDiv, Multiple Input-Single OutputMISO) and single antenna (Single Input Multiple Output-SIMO) modes are also analyzed and compared at overall manner and at channel-specific manners.

A survey of classification accuracy using multifeatures and multi-kernels

The bag-of-words (BoW) model is used widely for image classification. In this model, the image-level representations are designed using BoW frameworks from local low-level features, therefore we introduce our local low-level feature, called the denseSBP feature, using for BoW. We will evaluate performance in classification when using this feature. To increase average precision, we combine denseSBP feature with other features using Multiple Kernel Learning (MKL). In this work, we also propose the method called the integrated method, that it based on using multi-features and multi-kernels in SVM classification to derive the best classification accuracy for each category of a dataset. We perform the comparative analysis about classification accuracies of the method using MKL and the integrated method on image benchmark datasets. The experimental results show comparable classification accuracies of proposal methods with the state-of-the-art methods.