Kousuke Imamura

A sub-mW MPEG-4 motion estimation processor core for mobile video application

This paper describes a sub-mW motion estimation processor core for MPEG-4 video encoding. It features a Gradient Descent Search algorithm whose computation power is only 7% of the conventional 1:4-subsampling search, producing higher picture quality. Another feature is an optimized SIMD datapath architecture to decrease a clock frequency and an operating voltage. It has been fabricated with CMOS 5-metal 0.18 um technology. The measured power consumption to process a QCIF 15 fps video is 0.4 mW under 0.85 MHz@1.0 V.

VLSI Architecture Study of a Real-Time Scalable Optical Flow Processor for Video Segmentation

An optical flow processor architecture is proposed. It offers accuracy and image-size scalability for video segmentation extraction. The Hierarchical Optical flow Estimation (HOE) algorithm [1] is optimized to provide an appropriate bit-length and iteration number to realize VLSI. The proposed processor architecture provides the following features. First, an algorithm-oriented data-path is introduced to execute all necessary processes of optical flow derivation allowing hardware cost minimization. The data-path is designed using 4-SIMD architecture, which enables high-throughput operation. Thereby, it achieves real-time optical flow derivation with 100% pixel density. Second, it has scalable architecture for higher accuracy and higher resolution. A third feature is the CMOS-process compatible on-chip 2-port DRAM for die-area reduction. The proposed processor has performance for CIF 30 fr/s with 189 MHz clock frequency. Its estimated core size is 6.02 x 5.33 mm 2 with six-metal 90-nm CMOS technology.

Image Content Detection Method Using Correlation Coefficient between Pixel Value Histograms

An extraction method for searching for unauthorized copies of an image on the Internet is required in image search to make use of digital watermarks. In this paper, we propose an efficient two-stage image search method for extraction of illegal copies of a target image. The first stage is a pre-search, which searches for candidate of illegal images by some simple method. The next stage is the main search, which extracts embedded copyright information from the candidate and decides whether the image is an illegal copy of the target. In addition, we propose a simple image search method which uses the correlation coefficient between pixel value histograms of images as a pre-search method. The proposed pre-search method is useful because the proposed method is possible to combine with the current extraction technologies of embedded information. The performance of the proposed pre-search method is evaluated through computer simulations.

Fast codeword search algorithm for ECVQ using hyperplane decision rule

Vector quantization is the process of encoding vector data as an index to a dictionary or codebook of representative vectors. One of the most serious problems for vector quantization is the high computational complexity involved in searching for the closest codeword through the codebook. Entropy-constrained vector quantization (ECVQ) codebook design based on empirical data involves an expensive training,phase in which Lagrangian cost measure has to be minimized over the set of codebook vectors. In this paper, we describe a new method allowing significant acceleration in codebook design process. This method has feature of using a suitable hyperplane to partition the codebook and image data. Experimental results are presented on image block data. These results show that our method performs better than previously known methods.

Automatic moving object extraction using x-means clustering

The present paper proposes an automatic extraction technique of moving objects using x-means clustering. The proposed technique is an extended k-means clustering and can determine the optimal number of clusters based on the Bayesian Information Criterion(BIC). In the proposed method, the feature points are extracted from a current frame, and x-means clustering classifies the feature points based on their estimated affine motion parameters. A label is assigned to the segmented region, which is obtained by morphological watershed, by voting for the feature point cluster in each region. The labeling result represents the moving object extraction. Experimental results reveal that the proposed method provides extraction results with the suitable object number.

Fast VQ encoding algorithms using angular constraint

In this paper, we propose two fast encoding algorithms for vector quantization. The first algorithm uses three significant features of a vector, that is , the norm, its projection angle to a reference line and the mean, to reduce the search area and accelerate the search process. The second algorithm has feature of using a suitable hyperplane to partition the codebook and image data. These algorithms allow significant acceleration in the encoding process. Experimental results are presented on image block data. These results confirm the effectiveness of our proposed algorithms.

Study on Seam Carving for Image Fingerprint

In past several years many image contents are distributed via internet. But illegal copy problem are rise up. Multimedia fingerprint techniques are tools for resist illegal copy. However today collusion attack such as average attack is appeared. Collusion attack is efficient method for breaking copyright protection scheme of image fingerprint. This paper presented an image fingerprint scheme to resistant collusion attack which is average attack especially. Proposed image fingerprint scheme is applied seam carving technique. Experimental results show that proposed image fingerprint scheme was good performance.

Watershed algorithm for moving object extraction considering energy minimization by snakes

MPEG-4, which is a video coding standard, supports object-based functionalities for high efficiency coding. MPEG-7, a multimedia content description interface, handles the object data in, for example, retrieval and/or editing systems. Therefore, extraction of semantic video objects is an indispensable tool that benefits these newly developed schemes. In the present paper, we propose a technique that extracts the shape of moving objects by combining snakes and watershed algorithm. The proposed method comprises two steps. In the first step, snakes extract contours of moving objects as a result of the minimization of an energy function. In the second step, the conditional watershed algorithm extracts contours from a topographical surface including a new function term. This function term is introduced to improve the estimated contours considering boundaries of moving objects obtained by snakes. The efficiency of the proposed approach in moving object extraction is demonstrated through computer simulations.

Matching pursuits using slant patterns and its dictionary design [video coding applications]

Video coding methods have been based on hybrid coding which employ motion compensation and discrete cosine transforms. But another highly efficient video coding method, at low bit rates, using matching pursuits, is another waveform coding technique which has been suggested recently. The matching pursuits algorithm represents a signal approximately using a dictionary, therefore its performance depends heavily on the dictionary. Matching pursuits generally use a separable dictionary composed of one dimensional basis functions. But, it is difficult for a separable dictionary to code images including slant patterns efficiently. In this paper, we design the dictionary for slant patterns by a simple geometric transform. And we propose a dictionary design algorithm by employing a learning technique for the required memory and complexity reduction. We evaluate the performance of the designed dictionary by computer simulations.

Mosaic-like reconfiguration scheme of video for resistance to average value attack

Recent years have witnessed a drastic rise in the number of video and image hosting sites, which provide and distribute image and video content on the Internet. In this context, several digital watermarking techniques have been investigated in order to prevent illegal duplication and distribution of such content. Illegal users in turn attempt to remove digital watermarks from such content, and collusion attacking methods, such as an average value attack, constitute a serious threat to digital watermarks. In previous research, we proposed a mosaic-like reconfiguration scheme for static images to resist average value attacks. The reconfiguration scheme comprises two components: a line clipping part that eliminates seam lines from an original image signal, and a line adding part that adds seam lines to the image signals of the preceding part. In this paper, we adapt this scheme to video signals.