Video summarization is one of the most efficient methods for retrieval from large video archives. An automated method is described for generating TV program trailers (short video clips to advertise the program). Our method uses introductory text from an electronic program guide, which is a short description of the program highlights. We assume that the video segments corresponding to closed caption sentences with similar expressions to those of the introductory text contain appealing scenes, and our system generates a program trailer by linking these segments together. The AdaBoost algorithm is used to learn the textual characteristics of the introductory text. The proposed method was used to summarize actual TV programs to test its effectiveness.
We developed an image display system which turns the back light of an LCD into a laser light source, and 2-dimensionally scans the back light. The system consists of a liquid crystal panel, a laser light source, a laser light source power control, and a 2-dimensional scanning mechanism. We achieved highly efficient image display properties shown below by this system. The system expanded the color reproduction region to 189%, shortened the Moving Picture Response Time to one half, and improved the contrast ratio to 6753:1.
We propose a color reproduction method without an infrared cut filter (IRCF) for digital color cameras. The infrared cut filter in a camera eliminates infrared rays over 700 nm in order to adjust the spectrum sensitivity of the image sensor to the color-matching functions of the human eye. Without an IRCF, the spectrum sensitivity of the image sensor fails to satisfy the Luther condition. To achieve good color reproduction without an IRCF, we employed two techniques (: 1) a color correction matrix with nonlinear terms and (2) changes to the matrix coefficient according to the color temperature of illumination. Results demonstrate that our techniques enable a better color reproduction than conventional methods without an IRCF.
We propose a novel method for 3d modeling using a fusion of a laser range sensor, a camera, and a flashlight. This combination provides dense normals and surface colors that can be mapped on a 3d model, whereas conventional sensors only output point clouds of the 3d geometry. Furthermore, the fusion enables formulations to be made simply and practically. Multi-view photometric stereo is used for estimating the fine normal distribution with a basic shape measured by the laser range sensor. Our photometric stereo can easily handle near-light formulation and specularity. Detailed surfaces can be shown by applying the normal map as bump mapping to the basic shape. Robust estimation and clustering are used for estimating reflection parameters. Results demonstrate that our method can estimate highly accurate reflection parameters and provide fine surface appearances using only a small amount of data. The effectiveness of our method is shown with an application of 3d content.
Recently, various Internet businesses have been expanding. Those that have adopted the business model combining both store and Internet sales are especially attracting attention. It is said that this "click-and-mortar" business model obtains the synergy effect. This research aims to extract features of the store and Internet sales. Furthermore, on the basis of the extracted features, we seek to find a marketing method in the hope of expanding the click-and-mortar synergy effect.
National Oceanic and Atmospheric Administration (NOAA) and Advanced Very High Resolution Radiometer (AVHRR) data are available on a daily basis and have been frequently used for global observation. Geometric correction is very important as a preprocessing technique for long-term analysis or monitoring that makes use of the data. Ground control points (GCPs) are set to manually determine the geometric transform of the equation because 4-10 pixels of error can arise in the geometric correction based on only the system information. Therefore, we propose an algorithm that automatically extracts GCPs with information of run-length matrix and on adjacent areas. The proposed method has two steps: the creation of measurement images and measurements of the run-length matrix. Results suggest that the proposed approach is suitable for extracting GCPs from NOAA-AVHRR data.