Rapid filtering of large medical images using one-dimensional convolution kernels.

Many image enhancement algorithms can be implemented by convoluting the image with a predefined kernel function. A class of such functions is identified which has the property that a two-dimensional convolution can be exactly implemented using two successive one-dimensional convolutions along the rows and columns of the image. This leads to a simple and rapid algorithm for image enhancement, which can be applied to implement both low pass and high pass filters. The technique enables large convolution masks to be used without excessive computation time and without the need for large intermediate storage arrays. This approach facilitates the interactive processing of high resolution digital medical images.