摘要
针对栅格图像中稠密坐标网格是影响矢量化效果和速度的重要原因,提出了先基于±45°线检测模板来快速剔除低于所选阈值的坐标网格,然后依据坐标网格剔除后的当前图像与过滤后的图像差异,识别曲线的端点。并逐一地沿各端点的走向进行扫描识别和重建被连带去除的曲线像素。另外,对某些图像瑕疵的影响而不能识别的曲线端点,采用基于计算机技术来选取曲线缺口区域,并沿区域内各点的走向进行识别和重建被连带去除的曲线像素。实验显示所提出的算法对剔除稠密坐标网格具有较好的效果。
Due to that dense coordinate grid in raster image is an important cause influencing the effect and velocity of vectorization, we proposed to fast eliminate first the coordinate grid with greyscale less than chosen threshold value according to line detecting templates with ＋45~ and then to recognize curve endpoints by the difference between current image with coordinate grid eliminated and the filtered image. Curve pixels eliminated jointly are scanned and recognized along the change directions of each curve endpoint and to be reconstructed. For those curve endpoints can＇t be recognized due to some image blemishes, curve gap region would be chosen by computer technique and the jointly eliminated curve pixels can be recognised by scanning along the change directions of every point in the region and then to be recon- structed. Experiment shows that the above algorithm has better effect on eliminating dense coordinate grid.
出处
《计算机应用与软件》
CSCD
2010年第8期263-265,288,共4页
Computer Applications and Software
关键词
稠密坐标网格
±45°线检测
网格剔除
被连带除去的曲线像素识别与重建
Dense coordinate grid Line detecting with ±45° Elimination of the grid Recognition and reconstruction of curve pixels jointly eliminated
