Development of an Intelligent Image Processing System by using Soft Computing Methods
Mario Köppen
TU Berlin, 2005
Abstract As a consequence of the increasing requirements from their operational areas, modern image processing systems (IPS) exhibit more and more increasing complexity. Manual adjustments for achieving optimal system performance are no longer sufficient. The modern adaptive procedures of Soft Computing makes it possible to obtain an optimal adjustment for the later employment already in the process of developping an IPS. The approach presented in this thesis work exhibits the development of a flexible, adaptive system for the efficient configuration of an IPS, fulfilling the following criteria
- high system performance,
- high system throughput,
- high flexibility of the organization and
- high degree of autonomy
and is considered here as the crucial activity of the developer of such a system. By considering the theoretical basis of Soft Computing, opportunities for the practical applications will be shown. In addition, the limitations of the employment of these methods are pointed out. The implications for the development of image processing systems within these theoretical borders are represented in the form of design criteria. By the example of a system for the optimal adjustment of texture filters, exemplarily a system approach based on such a design is presented. The results from the employment of this system confirm the practical usability of the provided solution.
Download the thesis as PDF file (sorry, only available in German): Thesis