基于视觉的拓扑特征提取

Topology based on Visual features extraction

摘要:

根据拓扑性质知觉原理,视觉原语是从整体到局部,拓扑感知优先。根据此仿生原理,建立感知目标物的形状拓扑信息的数学模型。首先提取物体三维轮廓信息;将空间图形轮廓降维映射到二维投影平面中进行处理；建立提取空间图形拓扑特征的数学模型。将轮廓拓扑空间同胚映射到(Hausdorf或度量空间或)二维投影平面,保持了拓扑不变性,因此便于分析出拓扑特征,可继续应用拓扑映射原理,分析点集拓扑。最终,拓扑特征与知识库模型相匹配,识别出物体。

Abstract:

According to Chen Lin"s theory of topological perception, Visual primitives from the whole to the part, topology-aware is preferred.This study based on Bionic principles, perceptions shape the object topology information. First three dimensional contour extraction information in addition, graphic map outline dimension to a two-dimensional projection plane processing; establishment of mathematical model for extraction of topological characteristics of spatial pattern.Topological space homeomorphic maps to the outline (Hausdorf, or metric space or) two-dimensional projection plane, keeping topological invariance, thus facilitating analysis of topological characteristics may continue to apply principle of topological maps and analysis of point set topology. Eventually, topological features match the repository model, identify objects.