Image Compression Based on Artificial Intelligent Techniques
AL-Rafidain Journal of Computer Sciences and Mathematics,
2009, Volume 6, Issue 3, Pages 75-109
AbstractThis research present four methods to compress digital images using clustering based on artificial intelligent techniques that include neural network, fuzzy logic and hybrid between them. To enhance the performance of the compression system, the first method was developed in two types (k-means 1 dimension run length encoding km1D, k-means 2 dimension run length encoding km2D) by applying traditional clustering algorithm k-means on color and gray level images and then apply compression algorithm RLE in one and two dimension by zigzag scanning to obtain compressed image. The second method (fuzzy c-mean 1dimension run length encoding fcm1D, fuzzy c-mean 2dimension run length encoding fcm2D) used fuzzy c-mean to apply clustering operation and then compression. The third method (kohonen 1 dimension run length encoding Koh1D, kohonen 2dimension run length encoding Koh2D) used kohonen neural network for clustering image and then used RLE. The fourth developed method (fuzzy kohonen 1dimension run length encoding fKoh1D, fuzzy kohonen 2dimension run length encoding fKoh2D) based on hybrid kohonen neural network and fuzzy logic i.e fuzzy kohonen network which is recognized as the best method among the four methods. The four compression methods that are implemented in this research are efficient when applied on gray level and color images.
- Article View: 64
- PDF Download: 172