Generating Mosaic Images Based On Texture Analysis

Photo-mosaic (mosaic-image) generation is the process of dividing an input image into equal rectangular blocks (image blocks), each of which is replaced with another image (tile image) that matches the features of a corresponding image block. When the produced photo mosaic is seen from a distance, it seemingly forms the input image.<br /> Photo mosaic is not a new concept; however, only a few publications on this subject are available because of its commercial nature. The quality of the produced photo mosaic depends on the size of the database and the variety of images. In addition to process time, a bottleneck occurs because of the discrepancy in the match rate between the input image and the produced photo mosaic.<br /> This research proposes three intelligence-based approaches for producing photo mosaics in less time: (1) k-means clustering with Manhattan distance, (2) back propagation neural network with Manhattan distance, and (3) hybrid fuzzy logic with Elman neural network.<br /> Three groups of features are extracted and then used to find the best matching tile image for each corresponding image block within the container image. The first group comprises statistical features extracted from a 64-gray level quantized histogram. These features are variance, mean, skewness, kurtosis, and energy. The second group is a subset of Tamura features, namely, coarseness, contrast, and directionality. The last group includes a feature called edge rate, which is computed as the percentage of edge pixels within an image that can be detected using a Canny filter. In addition, two methods of color correction are used to adjust the colors of a tile image to match the colors of a corresponding image block. The first method is based on mean, whereas the second is based on histogram specification.<br /> Finally, experimental results show that hybrid fuzzy logic with Elman neural network is the best among the three approaches used. This technique needs 10.0 seconds to produce photo mosaics, with a correlation rate of 0.82 between the container image and the produced photo mosaics,using mean-based color correction. By contrast, this approach needs 42.33 seconds to produce photo mosaics, with a correlation rate of 0.86, using histogram specification-based color correction.