Stavros Paschalakis

Double precision floating-point arithmetic on FPGAs

We present low cost FPGA floating-point arithmetic circuits for all the common operations, i.e. addition/subtraction, multiplication, division and square root. Such circuits can be extremely useful in the FPGA implementation of complex systems that benefit from the reprogrammability and parallelism of the FPGA device but also require a general purpose arithmetic unit. While previous work has considered circuits for low precision floating-point formats, we consider the implementation of 64-bit double precision circuits that also provide rounding and exception handling.

A low cost FPGA system for high speed face detection and tracking

We present an FPGA face detection and tracking system for audiovisual communications, with a particular focus on mobile videoconferencing. The advantages of deploying such a technology in a mobile handset are many, including face stabilisation, reduced bit rate, and higher quality video on practical display sizes. Most face detection methods, however, assume at least modest general purpose processing capabilities, making them inappropriate for real-time applications, especially for power-limited devices, as well as modest custom hardware implementations. We present a method which achieves a very high detection and tracking performance and, at the same time, entails a significantly reduced computational complexity, allowing real-time implementations on custom hardware or simple microprocessors. We then propose on FPGA implementation which entails very low logic and memory costs and achieves extremely high processing rates at very low clock speeds.

Real-time face detection and tracking for mobile videoconferencing

This paper addresses the issue of face detection and tracking in the context of a mobile videoconferencing application. While the integration of such technology into a mobile videophone is advantageous, allowing face stabilization, reduced bandwidth requirements and smaller display sizes, its deployment in such an environment may not be straightforward, since most face detection methods reported in the literature assume at least modest processing capabilities and memory and, usually, floating-point capabilities. The face detection and tracking method which is presented here achieves high performance, robustness to illumination variations and geometric changes, such as viewpoint and scale changes, and at the same time entails a significantly reduced computational complexity. Our method requires only integer operations and very small amounts of memory, of the order of a few hundred bytes, facilitating a real-time implementation on small microprocessors or custom hardware. In this context, this paper will also examine an FPGA implementation of the proposed algorithmic framework which, as will be seen, achieves extremely high frame processing rates at low clock speeds.

The MPEG-7 Video Signature Tools for Content Identification

This paper presents the core technologies of the video signature tools recently standardized by ISO/IEC Moving Picture Experts Group (MPEG) as an amendment to the MPEG-7 Standard (ISO/IEC 15938). The video signature is a high-performance content fingerprint that is suitable for desktop scale to web-scale deployment and provides high levels of robustness to common video editing operations and high temporal localization accuracy at extremely low false alarm rates, achieving a detection rate in the order of 96% at a false alarm rate in the order of five false matches per million comparisons. The applications of the video signature are numerous and include rights management and monetization, distribution management, usage monitoring, metadata association, and corporate or personal database management. In this paper, we review the prior work in the field, explain the standardization process and status, and provide details and evaluation results for the video signature tools.

Recent developments on standardisation of MPEG-7 Visual Signature Tools

This paper presents the latest developments and possible new directions for future work in standardisation of Visual Signature Tools within the Moving Picture Experts Group (MPEG). The tools, which include the Image Signature descriptor and the recently completed Video Signature descriptor, form a part of the MPEG-7 specification. They enable fast and robust detection of duplicate or derived visual media content, images and videos. Descriptors of this type are sometimes also referred to as fingerprints or robust hashes. Here we mainly focus on introducing the technology behind the recently completed Video Signature Tools and describe some recent developments and demonstration applications for the Image Signature Tools. Finally, we briefly present MPEG exploratory investigations on requirements of searching for different images containing the same visual objects within the mobile visual search framework.

Combined geometric transformation and illumination invariant object recognition in RGB color images

This paper presents a novel approach for object recognition in RGB color images using features based on the theories of geometric and complex moments. By effectively combining the properties of the RGB color space and the normalization procedures and properties of the geometric and complex moments we have implemented a feature vector that is invariant to geometric transformations (i.e. translation, rotation and scale) and changes in both the illumination color and illumination intensity. The experimental results presented demonstrates the performance of the proposed feature set and investigate its tolerance to image distortions.

An FPGA System for the High Speed Extraction, Normalization and Classification of Moment Descriptors

We propose a new FPGA system for the high speed extraction, normalization and classification of moment descriptors. Moments are extensively used in computer vision, most recently in the MPEG-7 standard for the region shape descriptor. The computational complexity of such methods has been partially addressed by the proposal of custom hardware architectures for the fast computation of moments. However, a complete system for the extraction, normalization and classification of moment descriptors has not yet been suggested. Our system is a hybrid, relying partly on a very fast parallel processing structure and partly on a custom built, low cost, reprogrammable processing unit. Within the latter, we also propose FPGA circuits for low cost double precision floating-point arithmetic. Our system achieves the extraction and classification of invariant descriptors for hundreds or even thousands of intensity or color images per second and is ideal for high speed and/or volume applications.

Low-cost accurate skin detection using color and pseudo-morphological features

This paper presents a new method for accurate skin detection. Skin detection techniques are used in various image and video processing applications, such as image categorization, face detection and tracking and, more recently, in selective image enhancement for digital TV products. The proposed method achieves high accuracy using a combination of color information and pseudo-morphological processing, and is suitable for low-cost and custom hardware implementations.

Compressed signature for video identification

This paper presents a new application-specific lossless compression scheme developed for video identification descriptors, also known as video fingerprints or signatures. In designing such a descriptor, one usually has to balance the descriptor size against discriminating power and temporal localisation performance. The proposed compression scheme alleviates this problem by efficiently exploiting the temporal redundancies present in the video fingerprint, allowing highly accurate fingerprints which also entail low transmission and storage costs. In this paper we provide a detailed description of our compression scheme and a comparative evaluation against well known state-of-the-art generic compression tools.

Low-cost hierarchical video segmentation for consumer electronics applications

This paper presents a new technique for the hierarchical segmentation of video content. Such a segmentation allows users to quickly and efficiently navigate a video, for example to find a specific point, skip unwanted parts, skim parts previously viewed, and so on. The proposed method achieves high segmentation accuracy at a low computational cost, making it suitable for implementation on mainstream consumer electronics.