基于颜色与线段的图像轮廓提取算法

传统二维图像轮廓识别算法通常是在图像中找到边缘,根据设定的条件将边缘像素组合起来形成轮廓。针对边缘检测算法通常需要对不同的图像设定不同的阈值,对不同类型的图像很难找出统一的阈值的问题,文中提出了一种同时利用图像中颜色与线段信息的彩色图像轮廓提取算法,算法采用自顶向下的颜色空间融合和自底向上的线段检测的方法,在初步获取边缘信息之后,综合利用检测结果生成目标的轮廓。算法的优点在于不需要进行阈值调整,实验表明:该方法可以有效地提取彩色图像中的目标轮廓。

Image Contour Extraction Algorithm based on Color and Line Segment Detection

Traditional edge detection algorithms always need to set different thresholds for different types of image, it＇s relatively difficult to find a suitable threshold once for all. Thus the top-down fusion of color space and the bottom- up line segment detection are adopted to obtain primitive edge pixels simultaneously, and then these pixel sets are combined to achieve the final contour. This algorithm requires no parameter tunings. Experiments indicate that this algorithm could effectively extract the object contour in the colored image.