Application of a transparent interface methodology to image processing

Many software engineering projects involve a significant design component in which an algorithm must be formulated as a sequence of processing steps that meets a solution criterion. As the problems tackled become more complex, it becomes increasingly important to create and use tools that help designers understand and manage the design process. We demonstrate the use of design tool called T-STAR in the domain of image processing, in which a toolkit called the TRAnsparent Image Problem Solving Environment (TRAIPSE) is extended to solve face-recognition problems. Key features of TRAIPSE are its visual interface to the space of partial image processing algorithms and its support for automatic assistance in exploring the space. The specific application we present is the analysis of human face images. (c) 2007 Elsevier Ltd. All rights reserved.