28-03-2012, 04:24 PM
A Low-Cost VLSI Implementation for Efficient Removal of Impulse Noise
New Microsoft PowerPoint Presentation.pptx (Size: 829.19 KB / Downloads: 96)
Impulse noise are removed using MEDIAN FILTER.
Affects noise-free pixels in the same ratio.
Image acquisition and transmission , noise corrupts the image.
Ultimate result : Image is blurred with increasing size of the filter and edges are distorted.
LOWER COMPLEXITY TECHNIQUES
Lower implementation cost.
Uses fixed size local window.
Few line Buffers should be involved.
Computational Complexity to be low same as the case of primary median filter.
The modification of this filter was introduced by Zhang & Karim i.e. DIFFERENTIAL RANK IMPULSE DETECTOR.
Taking this criteria under account a cost effective technique is Simple Edge-Preserved Denoising Technique
(SEPD)
SIMPLE EDGE-PRESERVED DENOISING TECHNIQUE (SEPD)
Storage space needed is 2 line buffers.
Simple arithmetic Operation are used such as addition and subtraction.
Edges are preserved although they might be impulse pixel points.
SEPD is composed of mainly three components-
EXTREME DATA DETECTOR
EDGE ORIENTED NOISE FILTER
IMPULSE ARBITER
VLSI IMPLEMENTATION OF SEPD
Needs 2 line Buffers so cost is low.
For better timing performance timeline structure is implemented so output can be produced at every clock cycle.
Since the operation of SRAM access belongs to the first pipeline stage of our design, we divide the remaining denoising steps into 6 pipeline stages evenly to keep the propagation delay of each pipeline stage around 6 ns.
The Architecture of SEPD consists of
Line Buffer
Register Bank
Extreme Data Detector
Edge oriented Noise filter
Impulse Arbiter
CONCLUSION
The extensive experimental results demonstrate that this design achieves excellent performance in terms of quantitative evaluation and visual quality, even when the noise ratio is as high as 90%.
As the outcome demonstrated SEPD outperforms other chips with the lowest hardware cost.
The architecture work with monochromatic images, but they can be extended for working with RGB color images and videos.