SUMMARYIn order to weave fabric based on a yarndyed fabric sample, the information on the thread arrangement of the fabric must be obtained. In this paper, we will discuss a method for automatically extracting the design elements of yarndyed fabric that correspond to thread placement information with the objective of automating a part of the fabric design process of yamdyed fabrics. In yamdyed fabrics, the pattern on the fabric's surface is forrned by the placement of different colored warp and weft threads. Consequently, the features of the warp and weft threads are always continuous in the warp and weft directions. As a result, we show that the bandpass image obtained by filtering the fabric's surface in the spatial frequency domain with a rectangular filter restricted to the warp direction or weft direction extracts information related to the thread placement in that direction. In addition, a simple clustering preprocessing method that detects the brightness is proposed. Then it is verified that the design elements of yarndyed fabric can be extracted based on clustering where the color difference in the bandpass image is the distance. When extracting yamdyed fabric elements, these methods are not concerned with individual regions of the pattern on the fabric's surface or features of each thread but are only concerned with the placement of the colored threads that define the design elements of the yarndyed fabric. They are also provided with an efficiency not available in earlier methods.

The visibility and spatial-frequency components of a design are discussed as visual features of yarndyed fabric design. The aim is to introduce objective visual features that can be used for classification in a database of yarndyed fabric designs. First, the visually effective distances are measured for yarndyed fabric design elements that have also been subjectively evaluated. The visually effective distance expresses the perceived differences between designs in terms of visibility at different viewing distances. Next, low-pass filtered images of the yarndyed fabric designs are correlated with the original images in terms of the measurement of the yarndyed fabric's identification distance. The spatial frequency components are a visual feature of yarndyed fabric design related to visibility and expressed by the visually effective distance. In addition, the crosscorrelations between the original image and low-pass filtered images at successively greater low-pass spatial frequency ranges are investigated. Then the characteristics of the reconstructed images are compared for different designs. The results confirm the effectiveness of using spatial-frequency components as visual features. This method uses the visually effective distance to capture the subjective visual features of a yarndyed fabric design. By virtue of its correspondence with objective spatial frequency components, this method provides feature evaluation capabilities and objectivity not found in earlier methods.Key words: Visibility of fabric designs; colored designs of woven fabric; spatial-frequency analysis of colored patterns.

This paper describes our design policy and prototype data collection of RWC (Real World Computing Program) multimodal database. The database is intended for research and development on the integration of spoken language and visual information for human computer interactions. The interactions are supposed to use image recognition, image synthesis, speech recognition, and speech synthesis. Visual information also includes non-verbal communication such as interactions using hand gestures and facial expressions between human and a human-like CG (Computer Graphics) agent with a face and hands. Based on the experiments of interactions with these modes, specications of the database are discussed from the viewpoint of controlling the variability and cost for the collection.