PENGELOMPOKAN BLOK RANAH BERDASARKAN RATA-RATA DAN VARIANSI INTENSITAS PIXEL PADA PEMAMPATAN CITRA DENGAN TRANSFORMASI FRAKTAL

• Main Author: Munir, Rinaldi
• Format: Theses
• Language: Indonesia
• Online Access: https://digilib.itb.ac.id/gdl/view/4829
• Institution: Institut Teknologi Bandung

<b>Abstract:<p align=\"justify\"> <br /> A software for image compression using fractal transformation has been developed. This software is called PCFrak (Pemampatan Citra dengan Transformasi Fraktal - Image Compression using Fractal Transformation).<p align=\"justify\"> <br /> The basic principle of fractal transformation is searching for a portion of image (called a domain block) which is self-similar to a smaller block (called a range block) within the same image, then devising an affine transformation that map the domain block into the range block. A set of affine transformations can be saved using smaller number of bits (bytes), so we get a more compact file. Reconstruction of the original image consist of iterating a set of affine transformations from any initial image. The reconstructed image is almost similar to the. original image. This method of compression can be classified into lossy.<p align=\"justify\"> <br /> A disadvantage of fractal transformation is that it needs a long compression time, because every range block has to be compared to all of the domain blocks. Classifying domain blocks based on the average and variance of pixel values can reduce compression time. This classification produces 72 classes. A range block is compared only to domain blocks in the same class. This implementation shows that this classification can reduce compression time significantly and also produce satisfying compression quality. <p align=\"justify\"><br /> PCFrak is implemented using Borland C++ Builder 1.0`s compiler on Windows 95 operating system. PCFrak compresses only greyscale images.