Shimal Das ¹, Dibyendu Ghoshal ², Jhunu Debbarma ¹

Affiliations
1 Department of Computer Science& Engineering, Tripura Institute of Technology, Agartala, IN
2 Departments of Electronics & Communication Engineering, National Institute of Technology, Agartala, IN

Abstract

A study on fractal geometry based image compression has been carried out in Red-Green-Blue (RGB), Hue Saturation Intensity (HIS) and modified HSI colour space and a comparative study has been carried out on Lena and Pine images. The modified HSI scheme incorporates the concept of non linearity in the saturation of the colour in the image. The nonlinear model of HSI scheme is more realistic approach than the traditional HSI colour space. The output results have been obtained in the form of compressed image. In modified non-linear model, the processing speed and Peak Signal to Noise Ratio (PSNR) have been found to yield greater value compared to the other method but compression ratio has been obtained with lower values. The non-linear saturation, despite being a more realistic phenomenon, the self-similarity in the image may be lower and this may be considered as the major reason for lower compression ratio and higher processing speed.

Keywords

Fractal Compression, Hue Saturation Intensity (HSI) Space, Modified HSI Scheme, Red-Green-Blue (RGB) Space, Self-Similarity.

Full Text

   

Shimal Das ¹, Dibyendu Ghoshal ²

Affiliations
1 Department of Computer Science and Engineering, Tripura Institute of Technology, Agartala - 799009, Tripura, IN
2 Departments of Electronics and Communication Engineering, National Institute of Technology, Agartala - 799055, Tripura, IN

Abstract

Color Image compression has become an important technology to decrease memory space and to reduce transmission time. The present study provides comparative values between RGB and YCbCr color spaces with respect to PSNR, execution time and compression ratio. Fractal coding is a valuable method which utilizes for compression of color image. This method is most suitable for irregular shape of images. Long encoding times still remain as main disadvantage of this compression technique. A hybrid fractal coding technique based on YCbCr scheme and Discrete Cosine Transform (DCT) with quadtree and isosceles triangle segmentation is proposed in the present study. The study needs six phases, first, conversion of input RGB image to YCbCr scheme; second, the resulting image (YCbCr) is partitioned separately based on quadtree approach and isosceles triangle segmentation approach; third, DCT is applied to every block of the partition, fourth, scanning of every block value and zigzag manner is applied to prevent zero co-efficient; fifth, both quad tree and isosceles triangle segmentation approaches are applied separately for having partitioned resulting image; sixth, to compress the image, Run Length Encoding (RLE) approach is utilized. Test results demonstrate that proposed strategy gives higher compression ratio, PSNR value with high processing speed without degrading the image quality.

Keywords

Affine Transformation, Discrete Cosine Transform (DCT), Fractal Compression, Isosceles Triangle Segmentation, Iterated Function System (IFS), Partitioned Iterated Function System (PIFS), Quadtree Partition, Run Length Encoding (RLE), YCbCr Transform

Full Text