An Efficient Decision Tree Based Architecture forRandom Impulse Noise Removal in images

Images are often degraded by impulse noise in the procedures of image acquirement and broadcast. In this paper, we propose an efficient VLSI Architecture of a Decision Tree Based Denoising algorithm which can be used for the effective removal of random-valued impulse noise in images. To accomplish the low cost, a low-complexity VLSI design is proposed. The main components of the proposed design are a Decision-Tree-Based Impulse Noise Detector to detect the noisy pixels in the corrupted image and an Edge Preserving Noise Filter to reconstruct the corrupted pixels. The Decision Tree Based Method includes a binary tree of three stages for the detection of the noisy pixel. The reconstructed pixels are written adaptively to the reconstructed image after the correction by the noise filter. The main feature of the proposed design is that it keeps the unaffected pixels untouched and thereby reducing the blurring effect in the reconstructed image. It requires only low computational complexity and two line memory buffers. The hardware cost of the proposed design is very less and therefore, the design is very appropriate to be applied to many real time scenarios.