| dc.description.abstract | In recent years, color images occupy an extensive area of the information used in computer technology. Therefore, how to efficiently process color images becomes a demanding task. In this thesis, two algorithms for processing color images are proposed.
Most of digital cameras have many appealing features, such as auto focus, auto exposure, etc, which enable user to easily take good pictures under various shot conditions, users still have chances of getting backlight images. This paper presents a new algorithm for compensating exposure in the case of backlighting, regardless of the position of objects. To achieve this compensation, the fuzzy C-means algorithm is first used to extract features from a backlight image. Then these extracted features are input into an SOM-based fuzzy system to determine the amount of compensation. A set of 26 images were tested to illustrate the performance of the algorithm.
In the second part of the thesis, A region based color reduction algorithm is proposed. In this algorithm, a superposed 3D histogram is first calculated. Then the sorted histogram list will be fed into a region- growing-and-merging-algorithm to determine the number of quantized color for each region. By using the computed numbers, the K-means algorithm is employed to extract the palette colors. Several experimental and comparative results illustrate the performance of the proposed algorithm. | en_US |