Digital Image Processing Using Noise Removal Technique

You might also like

Download as pptx, pdf, or txt
Download as pptx, pdf, or txt
You are on page 1of 15

z

Digital Image Processing Using


Noise Removal Technique: A
Non-Linear Approach
z
Digital Image Processing Using Noise Removal
Technique: A Non-Linear Approach

 Published in: 2022 IEEE Asia-Pacific Conference on Computer


Science and Data Engineering (CSDE)

 Date of Conference: 20 December 2022

 Date Added to IEEE Xplore: 06 April 2023

 Conference Location: Gold Coast, Australia


z
Abstract

 Undesirable by-product of a captured image: Image Noise

 Denoising is a common component in digital image processing

 Importance of image processing in today's rapidly growing


technologies

 Exploration of different types of image noise: photo electronic


noise, impulse (salt & pepper) noise, and Gaussian noise

 Review of linear approaches to denoise the image using 2D


convolution filters and median nonlinear filters
z
Introduction to Digital Image Processing

 Manipulating images using various algorithms

 Purpose: Improve pictorial information for human interpretation

 Integral part of today's technology

 Key components: Denoising to remove image noise

 Techniques explored: Linear and non-linear approaches


z
Methodology - Literature Review

 Objective: Qualitative approach to image denoising


using various filters

 Review of research papers on digital image


processing

 Focus on image restoration techniques and


challenges

 Exploration of different types of image noise and their


effects

 Tools used: Python 3.7 - OpenCV IDE for image


processing
z
Digital Image Processing Phases

 Image Acquisition: Capturing images in digital format

 Pre-processing: Manipulating images for better recognition

 Restoration: Improving image appearance using mathematical models

 Feature Extraction: Extracting unique features for further processing

 Interpretation: Identifying key significance and understanding image


elements

 Compression: Techniques for reducing image file size and storage


z
Image Noise

 Definition: Random distribution of brightness and color


misinformation

 Impact: Reduces image quality and clarity

 Causes: Camera lens imperfections, environmental conditions,


image degradation
z
Types of Digital Image Noise

 Photo Electronic Noise: Caused by insufficient light and thermal


noise

 Impulse Noise (Salt and Pepper): Result of miss transmission


and sensor malfunction

 Gaussian Noise: Normal electronic amplifier noise with


independent pixel values
z
Photo Electronic Noise

 Definition: Noise caused by insufficient capture of photons by


camera sensors

 Types: Poisson noise in low light environments, thermal noise


from random motion of charged carriers

 Impact: White noise affecting all frequency channels


z
Impulse Noise (Salt and Pepper)

 Definition: Noise caused by miss transmission


and sensor malfunction

 Characteristics: Independently impacting pixel


values, resulting in dark and bright spots

 Example: Salt and pepper noise causing dark


pixels in bright regions and vice versa
z
Gaussian Noise

 Definition: Normal electronic amplifier noise with


independent pixel values

 Characteristics: Affects each signal intensity


independently

 Impact: Amplified noise in dark areas of images with


RGB color channels
z
Image Restoration Process

 Objective: Improve appearance of degraded images

 Method: Applying mathematical and probabilistic models to


compensate for lost information

 Techniques: Filtering to enhance and modify images, including


smoothing, sharpening, and colorization
z
Image Smoothing (Linear Filter)

 Technique: Using 2D convolution to smooth image


noise

 Process: Applying different kernel sizes to convolve


original noisy image

 Outcome: Blurred edges and noise reduction, but


some degradation still visible
z
Nonlinear Filtering - Median Filter

 Technique: Using median filter to remove noise degradation

 Process: Replacing center pixel with median value of


neighborhood pixels

 Kernel Sizes: 3x3, 5x5, or 7x7 for effective noise removal

 Advantage: Maintains sharpness of pixels without blurriness

 Effectiveness: Highly efficient in removing Gaussian and


impulse noise
z
Conclusion

 Summary: Overview of image noise and its impact on image quality

 Importance: Digital image processing plays a crucial role in


enhancing image interpretation

 Key Findings: Different types of image noise require specific


denoising techniques

 Future Directions: Opportunities for further research and


development in image denoising

 Conclusion: Nonlinear filtering, particularly median filtering, is


effective in removing noise while maintaining image sharpness.

You might also like