Kai-Lung Hua

Computer-aided classification of lung nodules on computed tomography images via deep learning technique

Lung cancer has a poor prognosis when not diagnosed early and unresectable lesions are present. The management of small lung nodules noted on computed tomography scan is controversial due to uncertain tumor characteristics. A conventional computer-aided diagnosis (CAD) scheme requires several image processing and pattern recognition steps to accomplish a quantitative tumor differentiation result. In such an ad hoc image analysis pipeline, every step depends heavily on the performance of the previous step. Accordingly, tuning of classification performance in a conventional CAD scheme is very complicated and arduous. Deep learning techniques, on the other hand, have the intrinsic advantage of an automatic exploitation feature and tuning of performance in a seamless fashion. In this study, we attempted to simplify the image analysis pipeline of conventional CAD with deep learning techniques. Specifically, we introduced models of a deep belief network and a convolutional neural network in the context of nodule classification in computed tomography images. Two baseline methods with feature computing steps were implemented for comparison. The experimental results suggest that deep learning methods could achieve better discriminative results and hold promise in the CAD application domain.

Joint trilateral filtering for depth map super-resolution

Depth map super-resolution is an emerging topic due to the increasing needs and applications using RGB-D sensors. Together with the color image, the corresponding range data provides additional information and makes visual analysis tasks more tractable. However, since the depth maps captured by such sensors are typically with limited resolution, it is preferable to enhance its resolution for improved recognition. In this paper, we present a novel joint trilateral filtering (JTF) algorithm for solving depth map super-resolution (SR) problems. Inspired by bilateral filtering, our JTF utilizes and preserves edge information from the associated high-resolution (HR) image by taking spatial and range information of local pixels. Our proposed further integrates local gradient information of the depth map when synthesizing its HR output, which alleviates textural artifacts like edge discontinuities. Quantitative and qualitative experimental results demonstrate the effectiveness and robustness of our approach over prior depth map upsampling works.

Intelligent deployment of UAVs in 5G heterogeneous communication environment for improved coverage

With hard requirements of high performance for the next generation mobile communication systems, especially 5G networks, coverage has been the crucial problem which requires the deployment of more stations by the service providers. However, this deployment of new stations is not cost effective and requires network replanning. This issue can easily be overcome by the use of Unmanned Aerial Vehicles (UAVs) in the existing communication system. Thus, considering this as a problem, an intelligent solution is presented for the accurate and efficient placement of the UAVs with respect to the demand areas resulting in the increase in the capacity and coverage of the wireless networks. The proposed approach utilizes the priority-wise dominance and the entropy approaches for providing solutions to the two problems considered in this paper, namely, Macro Base Station (MBS) decision problem and the cooperative UAV allocation problem. Finally, network bargaining is defined over these solutions to accurately map the UAVs to the desired areas resulting in the significant improvement of the network parameters, namely, throughput, per User Equipment (UE) capacity, 5th percentile spectral efficiency, network delays and guaranteed signal to interference plus noise ratio by 6.3%, 16.6%, 55.9%, 48.2%, and 36.99%, respectively in comparison with the existing approaches.

A comparative study of data fusion for RGB-D based visual recognition ☆

Abstract Data fusion from different modalities has been extensively studied for a better understanding of multimedia contents. On one hand, the emergence of new devices and decreasing storage costs cause growing amounts of data being collected. Though bigger data makes it easier to mine information, methods for big data analytics are not well investigated. On the other hand, new machine learning techniques, such as deep learning, have been shown to be one of the key elements in achieving state-of-the-art inference performances in a variety of applications. Therefore, some of the old questions in data fusion are in need to be addressed again for these new changes. These questions are: What is the most effective way to combine data for various modalities? Does the fusion method affect the performance with different classifiers? To answer these questions, in this paper, we present a comparative study for evaluating early and late fusion schemes with several types of SVM and deep learning classifiers on two challenging RGB-D based visual recognition tasks: hand gesture recognition and generic object recognition. The findings from this study provide useful policy and practical guidance for the development of visual recognition systems.

A novel multi-focus image fusion algorithm based on random walks

Abstract In multi-focus image fusion, the aim is to create a single image where the whole scene is focused by fusing multiple images captured with different focus distances. The fused image has greater depth of field than each of the input images. In this paper, we present a new method for multi-focus image fusion via random walks on graphs. The proposed method first evaluates the focus areas in a local sense and identifies nodes corresponding to consistency of nodes in a global sense. Several popular feature sets based on focus measure and color consistency are evaluated and employed to create a fully connected graph to model the global and local characteristics, respectively, of the random walks. The behavior of random walks on the graph is utilized to compute the weighting factor for each of the shallow depth-of-field input image. Experimental results show that the proposed method outperforms many state-of-the-art techniques in both subjective and objective image quality measures.

Extended Guided Filtering for Depth Map Upsampling

This extended guided filtering approach for depth map upsampling outperforms other state-of-the-art approaches by using a high-resolution color image as a guide and applying an onion-peeling filtering procedure that exploits local gradient information of depth images.

Clothing genre classification by exploiting the style elements

This paper presents a novel approach to automatically classify the upperwear genre from a full-body input image with no restrictions of model poses, image backgrounds, and image resolutions. Five style elements, that are crucial for clothing recognition, are identified based on the clothing design theory. The corresponding features of each of these style elements are also designed. We illustrate the effectiveness of our approach by showing that the proposed algorithm achieved overall precision of 92.04%, recall of 92.45%, and F score of 92.25% with 1,077 clothing images crawled from popular online stores.

Depth map super-resolution via Markov Random Fields without texture-copying artifacts

The use of time-of-flight sensors enables the record of full-frame depth maps at video frame rate, which benefits a variety of 3D image or video processing applications. However, such depth maps are typically corrupted by noise and with limited resolution. In this paper, we present a learning-based depth map super-resolution framework by solving a MRF labeling optimization problem. With the captured depth map and the associated high-resolution color image, our proposed method exhibits the capability of preserving the edges of range data while suppressing the artifacts of texture copying due to color discontinuities. Quantitative and qualitative experimental results demonstrate the effectiveness and robustness of our approach over prior depth map upsampling works.

An efficient scheduling algorithm for scalable video streaming over P2P networks

Abstract During recent years, the Internet has witnessed rapid advancement in peer-to-peer (P2P) media streaming. In these applications, an important issue has been the block scheduling problem, which deals with how each node requests the media data blocks from its neighbors. In most streaming systems, peers are likely to have heterogeneous upload/download bandwidths, leading to the fact that different peers probably perceive different streaming quality. Layered (or scalable) streaming in P2P networks has recently been proposed to address the heterogeneity of the network environment. In this paper, we propose a novel block scheduling scheme that is aimed to address the P2P layered video streaming. We define a soft priority function for each block to be requested by a node in accordance with the block’s significance for video playback. The priority function is unique in that it strikes good balance between different factors, which makes the priority of a block well represent the relative importance of the block over a wide variation of block size between different layers. The block scheduling problem is then transformed to an optimization problem that maximizes the priority sum of the delivered video blocks. We develop both centralized and distributed scheduling algorithms for the problem. Simulation of two popular scalability types has been conducted to evaluate the performance of the algorithms. The simulation results show that the proposed algorithm is effective in terms of bandwidth utilization and video quality.

A robust tracking algorithm for 3D hand gesture with rapid hand motion through deep learning

There are two main problems that make hand gesture tracking especially difficult. One is the great number of degrees of freedom of the hand and the other one is the rapid movements that we make in natural gestures. Algorithms based on minimizing an objective function, with a good initialization, typically obtain good accuracy at low frame rates. However, these methods are very dependent on the initialization point, and fast movements on the hand position or gesture, provokes a lost of track which are unable to recover. We present a method that uses deep learning to train a set of gestures (81 gestures), that will be used as a rough estimate of the hand pose and orientation. This will serve to a registration of non rigid model algorithm that will find the parameters of hand, even when temporal assumption of smooth movements of hands is violated. To evaluate our proposed algorithm, different experiments are performed with some real sequences recorded with Intel depth sensor to demonstrate the performance in a real scenario.