Home | About | Introduction of the IIEEJ | Call for Papers | Sponsors | Paper Submission | Registration | Invited Speech | Program | Presentation | Workshop Committee | Venue | Tours and Accomodation

April 18/2021 01:49   

Invited Speech



    "The Role of Movie and Telephone in the History of Communication Media",   Prof. Ryohei Nakatsu

    Recently various types of new communication media have emerged. Good examples are e-mails, blogs, and social networks such as Twitters and Facebook. There are various opinions and critics concerning these new media. Some are favorite of these media and some others are against them. However most of them only look at the surface communication phenomena initiated by these media. And only few of such opinions and critics are trying to look into the depth of the phenomena that lie in the background. In this presentation I try to investigate this recent trend based on the long history of human communications. And I want to show that this trend could be interpreted as a shift from logical communications to emotional communications. At the same time I want to clarify that this recent trend has its origin to the invention of movies and telephony in late 19th century.

    "Visual Quality Evaluation for Images and Videos",   Prof. Ngan King Ngi

    Visual quality assessment of images and videos has become important nowadays with the demand of high-definition and high-quality viewing experience. The standards bodies, i.e., ITU-T VCEG and MPEG, now require subjective tests to be performed on all proposals.
This talk will give an introduction on subjective and objective visual quality assessments, with focus on the latter. For subjective visual quality assessment, a short review will be given on the standard subjective evaluation procedure and the subjective score analysis methods. For objective visual quality assessment, its applications, classifications, and different techniques used in image and video quality metrics will be discussed. The general framework used by the classical HVS-model based image quality metrics (IQMs) will be briefly introduced first, followed by some new methodologies which do not rely on the HVS models, e.g., ones used in SSIM (Structural SIMilarity) and VIF (Visual Information Fidelity) quality measures. For video quality assessment, how temporal information is used to measure the perceptual video quality will be presented.
During the talk, our works on visual quality evaluation will be introduced, including our subjective quality HD video database; just noticeable distortion (JND) models; a practical image quality metric and its usage in image coding; image quality assessment based on decoupling detail losses and additive impairments and its extension to video quality metric; and finally a method to measure video temporal inconsistency.

    "Design of Image Quality Metric using Tchebichef Moments",   Prof. P.Raveendran
    Images are ubiquitous in modern communication, allowing us to transfer information in many different forms (animated and static) and of different content (humans, animals, places, objects) in different genres and communication venues. To increase efficiency, a number of processes - compression, transmission, enhancement etc- are used to transform images for easier transmission; however these cause distortions to occur. Therefore, there is a need for the assessment of the degree of deviance of the duplicated/compressed/enhanced image from the original. As human beings are the end users of the images and their evaluation on the acceptability of the reproduction for their needs is still the critical criterion, human subjective assessment of image quality is still regarded the touchstone or benchmark of quality. However, as has been well-documented in the literature, subjective assessment suffers from a number of limitations that renders it less than perfect. It is time-consuming, expensive and inconvenient.
The current research priority, therefore, has been with finding an objective metric that most closely approximates this subjective evaluation while averting at least its major pitfalls. The objective image quality metric (IQM) can be categorized into three types based on the availability of the reference image. The most popular approach is known as full-reference quality assessment (FRQA), where the quality of a test image is obtained based on the comparison with the reference image which is assumed to be perfect in quality. Unfortunately, as sometimes the reference image is only partially available or not available at all, reduced-reference and no-reference types of quality assessments will come into the play. One direction has been in the working out of mathematical models such as mean squared error (MSE), peak signal-to-noise ratio (PSNR), Peak MSE, Laplacian MSE, Minowski Error etc. While the chief advantages of these metrics are their simplicity and mathematical tractability, they do not correlate well with perceived quality measurement, possibly because they do not take cognizance of the characteristics of the Human Vision System (HVS).
Since the pioneering work on the introduction of orthogonal moments as potential pattern features by us in 2003, there has been a steady growth in utilizing these moments for various applications by the by research community. In this presentation, Tchebichef moments are used as image features in local quality index formulation and image content (block) classification. Tchebichef moments are chosen due to its desirable properties.

Last updated: November 09, 2018 16:14