LEAST MEAN SQUARE ERROR BASED BLOCK TRUNCATION CODING FOR IMAGE COMPRESSION
N. Nithiyanandam · 2014
Among the various spatial domain image compression techniques, Block Truncation Coding (BTC) is one of the methods which has the least computational complexity. But unfortunately, the bit rate obtained for a block size of 4x4 is 2 bits per pixel. When the block size is increased for obtaining higher compression ratio, and thereby lower bit rates for a band limited channel, the annoying blocking artifacts and the blurred edges dominate the image. In this research thesis, modification of the two tone Block Truncation Coding is done in order to improve the performance of the traditional BTC. The bit rate is further reduced by increasing the block size to 8x8, 16x16, 32x32 and 64x64. The parameters such as Peak Signal to Noise Ratio, Root Mean Square Error and Contrast are measured and it is found that the proposed methods of BTC are superior to the traditional BTC. The contrast of the image is enhanced and the computational complexity of the modified methods is kept minimal. Another method namely a “Least Mean Square Error Based Block Truncation Coding” is formulated. This method of BTC reduces the mean square error of the whole image to a minimum thereby improving the Peak Signal to Noise Ratio. This paves the way for a nearly error free and compressed transmission of the images through the communication channel. A comparative study of the performance of the Traditional BTC, modified methods of BTC and the Least Mean Square Error based BTC is done finally and the results are tabulated. Four sample gray scale images have been taken for analysis and it is found the modified methods and the Least Mean Square Error based BTC are superior to the Traditional BTC. This work can be further extended to colour images as well. The two tone BTC can be incorporated separately to the Red, Green and Blue (R, G, B) components of a colour image and further compression ratio can be achieved.