Abstract: Aiming at the problem that traditional vector recognition algorithm has and low recognition efficiency, this paper uses a new method to recognize the digitized curves which is based on sparse pixel traversal tracking fitting algorithm. The algorithm operations with orthogonal pixel tracking style to obtain key points data of graphics area, then the obtained points are used to make curve fitting to uniformly perform the recognition of line segments, circles and arcs, et al. Experimental results show that this algorithm is more efficient and effective compared with conventional methods, especially on the consumed time and space.

Key words: scanning image, digital curve, orthogonal pixel, tracking fitting, vectorization algorithm

摘要： 针对传统矢量化算法在识别扫描图时识别效率不高的问题，使用稀疏像素遍历的跟踪拟和算法进行扫描图数字曲线识别。该算法以改进的正交像素跟踪方式获取图元区域关键点数据，通过相应的拟和公式实现直线段、圆和圆弧等数字曲线的提取。实验结果表明，相比传统矢量化算法，该算法在时间复杂度和空间复杂度方面具有较大优势。

关键词: 扫描图, 数字曲线, 正交像素, 跟踪拟和, 矢量化算法

CLC Number: 

• TP393