灰度不变的图象插值算法

现有的图象插值算法会改变局部平均灰度，导致图象对比度下降。本文提出了一种新的保持平均灰度不变的图象插值算法。此算法的关键不同点在于不是让算法结果在采样点的值等于原采样值，而是在一个象素大小的单位面积上得到的平均值等于原采样值。插值结果C1连续，视觉上的平滑性接近于样条插值。将该方法用于图象几何变换可以避免灰度的损失。

Image Gray Level Invariant Interpolation Algorithm

Traditional image interpolation algorithms may change the local average gray level and lead to the decrease of image contrast. A new kind of interpolation algorithm is proposed, which keeps the average gray level invariant. A key difference is that this algorithm does not give the same value at the sampling point but gives the same average value in unit area of each pixel. The interpolation result is C1 continuous and looks almost as smooth as spline interpolation. This algorithm can help geometric transformation to avoid the loss of intensity of images.