Atsushi Hiroike

Visualization of Information Spaces to Retrieve and Browse Image Data

We have developed a user interface for similarity-based image retrieval, where the distribution of retrieved data in a high-dimensional feature space is represented as a dynamical scatter diagram of thumbnail images in a 3-dimensional visualization space and similarities between data are represented as sizes in the 3-dimensional space. Coordinate systems in the visualization space are obtained by statistical calculations on the distribution of feature vectors of retrieved images. Our system provides some different transformations from a high-dimensional feature space to a 3-dimensional space that give different coordinate systems to the visualization space. By changing the coordinates automatically at some intervals, a spatial-temporal pattern of the distribution of images is generated. Furthermore a hierarchical coordinate system that consists of some local coordinate systems based on key images can be defined in the visualization space. These methods can represent a large number of retrieved results in a way that users can grasp intuitively.

High-Speed Similarity-Based Image Retrieval with Data-Alignment Optimization Using Self-Organization Algorithm

In a retrieval system for vast amounts of image data, the primary storage cannot hold all image feature vectors because a huge data capacity would be required. Therefore, it is necessary to handle a slow-access secondary storage effectively as well as a first-access primary storage. In this paper, we propose a data-alignment optimization method in the secondary storage to access fast. Our idea is based on the premise that similar data have a high probability of simultaneous access in a similarity-based retrieval system. Then, we approximate the secondary storage by one dimensional alignment and allocates similar data to closer regions in the alignment. It is regarded as an optimization problem for a one-dimensional alignment that enables a fast retrieval by reducing a seek range. Through the experiments for 5 millions dataset, we confirmed that the data access time was reduced by 25.3% compared to no data-alignment optimization, which contributed significantly to speeding up the entire retrieval process.

Graphical representation for generating musical sequences

The author constructed a model for mapping between music and visual representation, and developed related software for controlling MIDI instruments. This software can design musical sequences by defining graphical shapes consisting of line segments representing notes in a sequence. The shapes are drawn and the sequences are played simultaneously. The pitch and the time value of a note are represented, respectively, by the direction and the length of the corresponding line, and the intensity and the timbre are represented by the color and the shape of the node of the line. Line drawing is executed sequentially or in parallel. If a shape is embedded into a line of another shape to create a complex pattern, the drawing is executed recursively.

Visualaization. Image Retrieval System Using Semantic Visualization of Retrieved Images.

We have developed a similarity-based image retrieval system that represents retrieved images as a scatter diagram a semantic space. One axis of the space shows the suitability of a keyword assigned to each image. The suitabilities are estimated by linearly transforming the image features of the retrieved data, and the coefficients of the transformation are determined by multiple regression analysis. First, a-user selects images as key images. Then the system retrieves images and calculates the estimated suitablities variance among the retrieved images for each keyword, and presents the user keywords that give the large variances. Next, the user selects a keyword from the given keywords for each axis. Finally, the system displays the retrieved images in the semantic space spanned by the two axes. The system enables the user to apply the center of gravity among the features of the key images to the retrieval. The system thus effectively assists the user in retrieving images by integrating the semantic visualization and the center of gravity retrieval. Testing of the keyword suitability estimation showed that the estimation precision became high by using keywords with the large suitabilites variances.