A Novel Image Compression Approach Using DTCWT And RNN Encoder

Limitations in storage capacity and communication bandwidth made researchers to work more on image compression. Many algorithms have been implemented to compress an image for better compression ratio. Recently the application of neural networks to compress image increasing gradually and achieved great success. The proposed work uses Recurrent Neural Networks (RNN) based encoding scheme in wavelet domain transform. Dual-Tree Complex Wavelet Transform (DTCWT) uses three level tree structured Discrete Wavelet Transform (DWT) used to get eight frequency subbands, the lower sub-band output is given to RNN for final encoding process. The experimental results show that the proposed work achieves better compression ratio for higher pixel values as compared to state-of-the algorithms. The Compression Ratio (CR) can be increased for lower pixel values by using more number of iterations for the neural networks and also by modifying the wavelet components.