THE PATCHWORK METHOD AND IMAGE PROCESSING
P. D. CHARLES, Calum A. Brown · Fractals · 1994
The patchwork method uses triangular patches to estimate surface area as a function of patch size. The patch area can be interpreted as a scale of observation. Using the patchwork method, one can identify crossover scales, or thresholds, which distinguish scales where a surface is best described by Euclidean or fractal geometries. A method is proposed for application of the patchwork method to video images of luminance intensity data. By recursively subdividing an image and applying the patchwork method, or by using a moving average and applying the patchwork method to a set of small subimages, the patchwork method can be applied for each pixel in the source image to obtain a new set of pixel intensities. Depending which characteristic is used, crossover values or level of detail at varying scales of observation, the resulting intensities highlight regions of fractal complexity or boundaries between image segments best described by Euclidean or fractal geometries. Often, the resulting image displays interesting properties and is useful in feature recognition, image segmentation, and image enhancement.