Designing vision chip architecture for man-machine interface based on saccadic eye movement

点击翻译按钮获取中文摘要

In this research project, vision chip architecture for tracking saccadic eye movement has been developed, and its applications to man-machine interface are discussed. As results of the project (1)vision chip architecture for detecting rapid eye movement in real-time is proposed, (2)high-precision eye gaze estimation algorithm which complements the rough estimation provided by vision chip is proposed, and (3)novel user interface paradigm using rapid eye movement measurement is discussed.20031 Parallel algorithm of low resolution vision chipParallel algorithm for tracking objects is designed based on software simulation.2 Development of high-precision eye gaze estimation algorithmHigh-precision eye gaze estimation based on very high resolution slow imaging sensor is studied. This slow algorithm compensates the precision of vision chip which are running very high speed.3 Designing vision chipSample chip is designed to measure the basic characteristics of LSI.2004-20051 Parallel algorithm of low resolution vision chipParallel algorithm for tracking eye movement is designed. The resolution of the chip was 100x100 at this stage2 Development of high-precision eye gaze estimation algorithmHigh precision pupil tracking algorithm is developed based on Extended Kalman-filter algorithm.3 Implementation of algorithm 1 and validation of the chipThe parallel algorithm is implemented on FPGA and the performance of the chip is 146 kframe/sec for 100x100 PE.