基于B－样条的图像缩放相对于传统图像缩放方法（最近邻点、双线性、双三次等）的优势性在诸多文献和实验中得到了印证。但基于插值的B－样条的传统方法一样，没有考虑信息损失问题，是一种次优方法。为此Unser提出正交B－样条，并从理论上证明了其最优性。由于正交B－样条的采样池数在阶数大于3时难于获取，Lee发展易于计算且保持了相当精度的斜投影B－样条。B－样条良好的连续性使图像的边缘在缩放过程中被模糊了。为了克服这一问题，提出了带锐度保持的斜投影B－样条缩放，在缩放过程中利用B－样条的递推性质对存在一阶或二阶微分过零（即有小边缘）的采样区间缩小其积分范围至边缘附近，从而达到保持锐度的目的。实验结果表明该方法在保持缩放图像的连续性、清晰度方面都具有良好的效果。

Image resizing plays an important role in image processing and the corresponding methods are widely used in various kinds of image processing software. Recently, oblique projection based B-Spline image resizing has been developed. Compared to conventional image resizing methods such as the nearest neighbor interpolation, bilinear interpolation, and bicubic spline interpolation, this method performs better. The resized image using oblique projection method has more continuity than using other methods.For the same reason, the resized image is smoothed at the same time, especially at the places where exist sharp edges. Thus, some important information lying in the area is lost. This result is unexpected in image processing. In this paper, a sharpness preserving method in the procedure of image resizing is proposed in order to overcome the shortcoming of oblique projection B-Spline image resizing mentioned above.In the new method, the range of integral of an output pixel is changed when there are sharp edges including in the corresponding source pixels,thus the sharpness of the output pixel can be preserved. Results show that the method performs better in both continuity and sharpness preserving.