手绘非规则军标图形的结构化识别方法

在基于草图的态势标绘系统中,线状和面状军标图形通常是依具体地形、战况、意图等标绘的,没有规则的形状,难以采用基于表观的方法进行识别。针对这类手绘非规则军标图形提出了一种结构化识别方法,基于非规则军标图形集的特点定义了9种图元和4种图元间空间结构关系,将图形表示为图元向量和图元结构关系矩阵,通过匹配图元向量和图元结构关系矩阵来识别图形类别,并估计未知图形与模板图形之间的图元对应关系。实验结果表明:本文方法能有效克服图形形状变化的影响,对手绘的常用非规则军标图形的类别和图元对应关系具有较高的识别正确率和识别速度。

Structural Recognition of Sketched Irregular Military Marking Symbols

Sketch-based military marking diagrams consist of many linear and regional symbols. Typically, these symbols are irregularly shaped because they are drawn according to specific terrain or battle scenario with freehand. Therefore, they are also called irregular symbols. Traditional appearance-based methods are not appropriate for recognizing these symbols. A structural recognition method for sketched irregular military marking symbols was proposed. Symbols were decomposed and represented by 9 types of primitives together with 4 types of spatial relationships among them. By matching the primitive counting vector and the primitive relationship matrix of the unknown symbol with those of the templates, the class label could be recognized, and their primitive correspondences could also be estimated. Experimental results demonstrate that the proposed method is effective and efficient for recognizing sketched irregular military marking symbols.