Image Compression Using Mapping Transform with Pixel Elimination

Abstract

     In today's world, digital image storage and transmission play an essential role,where images are mainly involved in data transfer. Digital images usually take large storage space and bandwidth for transmission, so image compression is important in data communication. This paper discusses a unique and novel lossy image compression approach. Exactly 50% of image pixels are encoded, and other 50% pixels are excluded. The method uses a block approach. Pixels of the block are transformed with a novel transform. Pixel nibbles are mapped as a single bit in a transform table generating more zeros, which helps achieve compression. Later, inverse transform is applied in reconstruction, and a single bit value from the table is remapped into pixel nibbles. With these nibbles, pixel values of the block are reconstructed without any loss. The average method is used in reconstruction of excluded pixels. This approach achieves better quality in reconstructed test images at lower PSNR values ranging from 33dB to 44dB. Compression ratio achieved is more than 2. Correctness ratio achieved by proposed method is more than 0.5.