Image ranking system

Abstract

Image processing is a very vast subject which involves inputting an image into a system and doing some processing over that image. Image processing is used in many applications including face detection, image-based retrieval systems, pattern recognition, image similarity, recognition and the like. Operations such as 'image segmentation', 'geometric transformations' and 'image editing' are usually the basis, or the initial phase, of more complex systems. One obvious field which is directly involved with image processing is that of photography. Photoshop, which is an editing software, uses algorithms which assist the photographer to enhance and manipulate digital images both automatically and manually. These include brightness, hue and saturation of the image, automatic exposure balance of overall picture, image resizing, grayscales and black and white images amongst many other functions. Successful attempts have been made in developing software to automatically adjust an image. A typical research area involves systems which automatically describe the objects in an image based on previous knowledge in a storage database. The medical field also makes good use of image processing techniques to reliably analzye tumors, fractured areas and faults in blood flow. The proposed system for this thesis makes use of such image processing concepts to categorize images according to a set of photographic rules; mainly focusing on color, picture composition and exposure. The scope behind this system is to gather basic rules of photography and assign a score for each. Eventually, the data set of images is put into a number of categories according to these scores and ranked. The motivation of this system derives from my passion for photography with the purpose of this software possibly being used by amateur photographers who are willing to learn and experiment with the basic rules of photography.