快速B样条二维细微边缘检测算法

传统的边缘算法无法得到令人满意的图像细微边缘检测结果。作为对传统边缘算法的补充，这里提出了用于图像特征提取的细微边缘检测算法。采用快速B样条转换实现快速图像插值可以提高运算速度。这种基于B样条函数的算法能快速计算出信号的一阶导数，通过搜索图像一阶导数最大值的位置确定边缘点的位置。实验结果表明图像经插值放大3倍以后可以获得满意的细微边缘检测结果。

Fast 2D Interpolatory Texture Extraction with B-splines

Conventional edge operators cannot extract fine texture in images satisfactorily. In this paper, an interpolatory edge operator is proposed to deal with image texture extraction as a supplement to ordinary edge operators. Fast image interpolation is implemented by means of fast B-spline transform. The algorithm detects edge points by seeking points with local maximum values of first derivatives of images. The first derivatives can be calculated quickly by subtraction based on B-splines. From experimental results, we reach the conclusion that satisfactory results of fine texture extraction can be obtained when images are expanded by three times with interpolation.