VIPS - a digital image processing algorithm development environment

Digital image processing involves using a computer to apply a sequence of mathematical operations to a numerical representation of an object*. The desired result may be, for example, the measurement of the length of a feature, an enhanced image for display or even a decision on whether an object meets certain specifications. The application of digital image processing techniques to any particular problem may be split into two broad phases2. The first phase is to determine the image processing algorithm, i.e. the sequence of mathematical operations required to achieve the desired result; the second phase is to develop appropriate hardware to implement the solution. In some applications, the system that is used to develop the algorithm may be used in the final implementation. Algorithm development is much like carpentry. All of the various image processing operations are the tools used by the algorithm development specialist to work on the input image, which represents the raw material. The main difficulty in algorithm development is that there is little or no theory that may be used to guide the selection of the best operation from several that may be suitable3. In practice, the operations are chosen heuristically from a large number of possible operations. This is done by trying one operation and, if it performs unsatisfactorily, trying another until the desired result is obtained. For this reason, an image processing system used for algorithm development must be highly inter-