Binary image scaling by piecewise polynomial interpolation

Presented here is an algorithm for scaling binary images based on piecewise polynomial interpolation. The algorithm defines a set of convolution kernels which can be used to scale the original image data by an arbitrary scaling factor and to reduce or remove the aliasing artifacts. The convolution kernels are derived from a surface geometry that is mathematically defined over the original data. This algorithm solves a quantization error problem which had prohibited practical applications of any polynomial as an interpolant for image scaling. Its microscopic behavior has been analyzed in a software simulation testbed. It can be applied for scaling binary images in the areas of facsimile imaging and font scaling. It has been fully tested and implemented in a commercial product.