视觉信息的群体动态时空编码和选择性注意机制

The visual information processing and encoding have been studied based on the.fundamental assumption of spatial-temporal coding in dynamic cell assemblies. The project was focused on three important research aspects: Firstly the theories and algorithms of visual perceptional grouping and shape similarity measuring have been developed. They were concerning moving object.tracking, image enhancement and segmentation, psychological contour, as well as pattern recognition and competition.Secondly the studies of emergent dynamics of dynamic cell assemblies, the interaction between the visual grouping and selective attention, as well as the.21 multiple-layer neural network modeling of visual grouping have been systematically carried out..Thirdly it has proposed neurobiology-inspired comprehensive computational.model for figure-ground discrimination and target tracking which consist of motion.detector array, saccadic and smooth pursuit eye movement module, oscillatory neural.network module, and selective attention module as well. It has found that fruit flies has also attention state. It is most interesting that it has explored a simple choice-making behavior in fruit flies facing competing visual cues at flight simulator.

以脑动态神经元集群时空编码为工作假设，探索视觉信息的编码，表达和加工。主要研究动态神经元集群自发涌现的动力学；视知觉组织与形状相似形度量（运动识别；场景分割；心理轮廓及模式识别与竞争）的理论和算法；视知觉组织和选择性注意机制的相互作用。目标是发展视知觉组织与神经编码全局理论及完整的神经计算模型，并探索对信息高科技领域的应用。