Abstract: In the present paper three heterogeneous architectures are discussed for complex systems, dedicated to real-time image processing and analysis. Usefulness and applicability of the solutions described here results from the fact that the systems are reconfigurable, and their final architecture can always be tailored to the specific needs of the specific problem (mainly related to the applications of the described systems in the fields of automation and robotics). In the paper authors describe the ultimate solution, found after long-time studies, as well as the evolution of the system architectures resulting from the technological progress, the increase of expectations from the real-time visual systems and practical experience gathered by the authors during the 15 years long period. Attention has been focused on the increasing flexibility of the reconfigurable hardware-software systems, which is a guarantee that the technology will keep up with the increasing requirements of the users. In particular it has been shown that the hardware systems based on latest large density reprogrammable devices are able to achieve many utilitarian features like object-oriented nature, scalability or adaptability, which has been up to now reserved for the software systems. Keywords: Image Processing, Configurable Computing, Field Programmable Gate Array, Hardware-Software Co-design, Heterogeneous Architecture. I. INTRODUCTION In the early 80-ties it has become obvious, taking into account the small computing power of the general purpose processors, that for the realization of image processing and analysis hardware acceleration is necessary. Extensive review of references discussing the studies and realizations from that period is contained in the papers [1]–[5]. 1 Rector of the AGH Technical University, Kraków, Poland, rtad@biocyb.ia.agh.edu.pl 2 Department of Automatics, Biocybernetic Lab, the AGH Technical University, Kraków, Poland, mago@biocyb.ia.agh.edu.pl 3 Department of Electronics, the AGH Technical University, Krakow, Poland, wiatr@uci.agh.edu.pl 4 Department of Automatics, Biocybernetic Lab, the AGH Technical University, Kraków, Poland, zibi@biocyb.ia.agh.edu.pl Hardware realization of reconfigurable real-time video systems is the subject of research in the Departments of Automatics and Electronics in the AGH Technical University, Kraków, Poland since 1986. During that period several studies have come into being, concerning the architecture of video systems adapted to the real-time operation (required by the planned applications in the problems of automation and robotics). For the research purposes concerning these applications three working systems for image processing and analysis have been constructed, which in every case have been based on the latest technology available at the time for solutions of dedicated electronic systems. The experience gathered during that period as well as several studies, both theoretical and practical, lead the authors to a conclusion that the field of solution concepts for the architectures of dedicated vision systems is much more stable than the rapidly varying (at the pace dictated by the progress of electron technology) realization possibilities of hardware solutions. Therefore, after finding that during the construction of several visual processing hardware systems an original methodology have emerged as a byproduct, the authors have decided to describe some elements of that methodology in the present work. The specific feature of this methodology is the use of dedicated hardware resources, controlled by software, for the realization of image processing algorithms, paralell nature of the applied processing, adaptation to the pipeline nature of the video signal, the possibility of hardware reconfiguration for the given algorithm executed at the moment, the ability to select the hardware resources depending on the specific nature of the algorithm being executed and several other unique features, described below during the presentations of particular solutions. The primary observation that has been made in the beginninig of the described research was the possibility of diversification of the computing resources, effectively carrying out the image preprocessing operations, depending on the implementation of the further stages of image analysis (see Fig.1) [1]. Therefore all the systems described here make use of heterogeneous architectures, considerably diversifying the method of hardware-software realization of particular stages of image processing and analysis. Now several characteristic element of that heterogeneity will be discussed. Reconfigurable Image Processing Architectures Research and Current State of Art at the AGH Technical University Ryszard Tadeusiewicz 1 Marek Gorgoñ 2 Kazimierz Wiatr 3 Zbigniew Mikrut 4