We propose a bio-inspired visual processing system for pattern segmentation which parallelizes the imformation and shows high-speed performance and flexible features. In the conventional model, a global element had functions to receive input from the entire network and feed back. In the proposed system, each oscillator unit shares the function and the global element can be regarded as a data field. As a result, we successfully enhanced the features as a parallel and distributed system. We verify the features by experiments with real implemention of analog circuit. We showed the result of pattern segmentation with it.