Nai-Chung Yang

A fast MPEG-7 dominant color extraction with new similarity measure for image retrieval

Dominant color descriptor (DCD) is one of the color descriptors proposed by MPEG-7 that has been extensively used for image retrieval. Among the color descriptors, DCD describes the salient color distributions in an image or a region of interest. DCD provides an effective, compact, and intuitive representation of colors presented in an image. In this paper, we will develop an efficient scheme for dominant color extraction. This approach significantly improves the efficiency of computation for dominant color extraction. In addition, we propose a modification for the MPEG-7 dissimilarity measure, which effectively improves the accuracy of perceptive similarity. Experimental results show that the proposed method achieves performance improvement not only in saving features extraction cost but also perceptually similar image retrieval.

Color-Based Image Salient Region Segmentation Using Novel Region Merging Strategy

In this paper, we propose a novel unsupervised algorithm for the segmentation of salient regions in color images. There are three phases in this algorithm. In the first phase, we use nonparametric density estimation to extract candidates of dominant colors in an image, which are then used for the quantization of the image. The label map of the quantized image forms initial regions of segmentation. In the second phase, we define salient region with two properties; i.e., conspicuous; compact and complete. According to the definition, two new parameters are proposed. One is called ldquoImportance indexrdquo, which is used to measure the importance of a region, and the other is called ldquoMerging likelihoodrdquo, which is utilized to measure the suitability of region merging. Initial regions are merged based on the two new parameters. In the third phase, a similarity check is performed to further merge the surviving regions. Experimental results show that the proposed method achieves excellent segmentation performance for most of our test images. In addition, the computation is very efficient.

An effective and flexible image enhancement algorithm in compressed domain

Images with JPEG format using discrete cosine transform (DCT) is selected for most popular compression standards. Enhancement in the compressed domain offers two major advantages, including low computational complexity and storage space. However, the compression is achieved by block-based transform; therefore, it is hard to enhance image globally. In this paper, the main issue is to develop a global enhancement method that effectively reduces the introduced blocking artifacts and achieves excellent visual quality of enhancement. We propose a combined DCT matrix representation, which consists of 8n × 8n pixel arrays, to enhance the global information on images for removing block artifacts in compressed-domain. In addition, we propose the multi-enhancement factors based on spatial frequency for image enhancement in compressed domain. From the simulation results, the proposed method achieves not only excellent improvement for image enhancement but also reducing the blocking artifacts significantly.

Kalman Filtering Based Motion Estimation for Video Coding

Video compression is a very efficient method for storage and transmission of digital video signal. The applications include multimedia transmission, teleconferencing, videophone, high-definition television (HDTV), CD-ROM storages, etc. The hybrid coding techniques based on predictive and transform coding are the most popular and adopted by many video coding standards such as MPEG-1/2/4 [1] and H.261/H.263/H.264 [2, 3], owing to its high compression efficiency. In the hybrid coding system, the motion compensation, first proposed by Netravali and Robbins in 1997, plays a key role from the view point of coding efficiency and implementation cost [4-11]. A generic hybrid video coder is depicted in Figure 1. Fig. 1. A generic hybrid motion compensated DCT video coder. DCT Quan-tization De-Quan-tization IDCT Motion Compensation Motion Estimation Frame Buffer Variable Length Coder

Global image enhancement in the DCT domain

The JPEG that applies discrete cosine transform (DCT) is one of the most popular image compression methods. Enhancement in the compressed domain offers two major advantages, including low computational complexity and storage space. To enhance an image in compression domain, the main issue is to develop an effective technique that limits the introduction of blocking artifacts. In this paper, we propose a new merging strategy to generate global DCT-like matrix, which achieve not only global enhancement but removing block artifacts in compressed-domain. In addition, we also propose multi-enhancement factors based on the distribution of frequency for fine tune image enhancement. From the simulation results, the proposed method provides an effective improvement for image enhancement and significant reduction in the introduction of block artifacts.

Image Recovery Based on Effective Image Completion

Exemplar-based inpainting algorithm is a model used for large region removal. It combines the advantages of both inpainting and texture synthesis. This algorithm usually requires large amounts of computational cost for priority calculation and matching iteratively. However, for relatively flat damaged areas, the simple interpolation can work well for image restoration. Inspired by the advantages of the color interpolation and the exemplar-based inpainting methods, in this paper, we propose an adaptive restored approach based on gradient-based analysis. The evaluation shows that our method can avoid blurring artifacts and preserving correct propagated structures in the faster way.

Motion Vector Re-estimation for Trans-coding Using Kalman Filter

Recently, many visual devices are applied to mobile phone, PDA or other products. These devices can receive real-time videos from wireless network or internet. In order to watch video programs, the video frame sizes receiving from various devices must be properly changed. Although many fast downscaling algorithms have been proposed, such as SA, AWA and MDN, their performances do not achieve high visual quality. In this paper, we propose a method to address this problem. Considering the correlation of motion characteristics of neighbor blocks in temporal and spatial domain, the method applies embedded Kalman filter to traditional fast downscaling algorithms. The experimental results indicate that the performance of the proposed method for fast downscaling videos provides an effective improvement.

Template-Based Scene Classification for Baseball Videos Using Efficient Playfield Segmentation

A robust scene-classification algorithm is able to provide the ground truth for video abstraction and high- level events extraction. In this paper, an efficient playfield segmentation using learning Vector Quantization (IVQ) is introduced, which is able to adapt to the variations of field colors in diverse baseball videos, and then we propose a reduced filed map feature that possesses field-class concept rather than low-level feature and it can also accelerate retrieval performance. Finally, a template-based learning algorithm is proposed for scene classification without shot detection or keyframe extraction in advance. Experiments with the inside and outside tests show that our method is capable of classifying various scenes reliably.

Image retrieval using point- and block-based visual vocabulary

With the rapidly evolving computer technologies, the multimedia and vision applications, such as visual recognition, scene modeling, image retrieval, and image categorization attract significant attention. The visual words, a collection of local features of images, can be used to represent image information. By considering the content homogeneity of visual words, we design a visual vocabulary which contains macro-based and micro-based corresponding to feature points and key blocks visual words, respectively. We also apply the proposed visual vocabulary and description scheme to construct an image retrieving system. The performance evaluation of the systems indicates that the proposed visual vocabulary achieves promising results.

Robust Player Tracking and Motion Trajectory Refinement for Broadcast Tennis Videos

Player tracking and trajectory detection play an important role in the content analysis of broadcast tennis videos. It is still a challenge because the player size is small and many noises and interference exist in a tennis court, and therefore often results in a detection failure. In this paper, we propose a robust technique for player tracking and trajectory modification using an adaptive Kalman filtering. Experimental results indicate that the proposed method improves the successful rate of player tracking significantly, especially for the upper court players.