Agent oriented vector quantization using self-organization realized in JAVA

This paper proposes a self-organizing clustering method to the quantization of image data which is implemented on an agent oriented concept. All implementation is written in the JAVA language since the distributed processing, total object oriented structure, platform independence and network orientation can be realized. The self-organizing clustering method is a useful method for evaluating and clustering the input data without a supervisor. The agent oriented model makes it possible to cluster the input image data distributedly and independently. Thus, using these techniques, parallel distributed processing can provide good quantization ability for input image data with complicated and structured characteristics. In addition to the development of this method, multi-rated communication between the different platforms on the network environment can be realized by parallel calculation on different machines separately.