Efficient and Enhanced High Throughput Image Denoising Using Chronical Fuzzy Set

Abstract

A basic and fundamental step in image processing for any type of applications is removing noise from a query image. The significant ideal property of a best and perfect image de-noising model is that to preserve edges and noise has to be removed entirely. An efficient fuzzy based filter integrated with modified rules set is implemented in this research work for noise reduction of images corrupted with at-most noise. Proposed design consists of two variant stages. The first stage produces a fuzzy derivative for whole eight different directions. Final stage uses these fuzzy derivatives to implement fuzzy smoothing by weighting the participation of neighbouring pixel values. Type 2 fuzzy structure distinguishes the noisy pixels in the satellite picture and transforms the picture into a binary picture, which is gone through the Adaptive Nonlocal Mean Filter (ANLMF) for the noise rectification. Ultimately, for the picture improvement, the kernel-based addition plan has to be carried out, which is done through the proposed filtering of fuzzy. The above whole process christened as chronic fuzzy system. In Proposed architecture, we present an energy efficient algorithm for making the system more robust and it is developed on a Cadence 90-nm technology. The energy per sample for 8-bit test pattern has been reduced by 64% and power consumption is reduced by 44% when compared to existing architectures.