pape@agcrr.bio.ns.ca (Andrew Pape) (06/15/91)
I am doing a project with the Environmental Marine Geology dept. of the Bedford Institute of Oceanography in Dartmouth, Canada. We are doing coastal geological/geographic research and are looking at coastal erosion in parts of Eastern Canada and the arctic. We have a series of aerial photographs taken along the coast of Newfoundland, and we want to scan them in and manipulate the images with Image Processing software. I'm fairly new with the world of Image Processing. We are using UNIX, Mac, and MS-DOS machines here, and are looking for suggestions of software packages or techniques that can do the job. We want to do the following with the bit-mapped images: Manipulate the images so that we can identify the coast-line (water level), and filter out all extraneous data (in-land features, water levels) so that only the coast remains in the image. This can be a complicated job, because the water line is not always really obvious when the land and water profiles are gradual. A cliff is really east to identify and manipulate with image processing software. Compare the images with scanned photographs from ten years previous with the eventual goal of identifing erosion over the ten years. Compensating for the geometric errors (skewing, rotation) in the photograph (if it was taken at an angle rather than perpendicular to the ground) by finding constant reference points in each of the two photographs (buildings, etc.) and "rubber sheeting" the image (stretching it with some sort of algorithm so that the reference points match). Comparing the coastline of the two images (with equal reference points) and and making a calculation of the difference. This could be like a subtraction operation that may result in a plot and numeric calculation of the difference between coastlines. Save and print the filtered images (containing only coatline and some reference feautures) and the plots that are a result of the subtraction of one image from the other. The fractal analysis may be a way of simplifying images for easy comparison. If we were to find the fractal dimension of the coastline, we could compare them numerically with the fractal data. I don't know very much at all about fractal analysis. Any suggestions? Thanks.