Downloaded 791 times
Uploaded byHaitham Ahmed
Image denoising
This document discusses image denoising techniques. It begins by defining image denoising as removing unwanted noise from an image to restore the original signal. It then discusses several types of noise like additive Gaussian noise, impulse noise, uniform noise, and periodic noise. For denoising, it covers spatial domain techniques like linear filters (mean, weighted mean), non-linear filters (median filter), and frequency domain techniques that apply a low-pass filter to the Fourier transform of the noisy image. The document provides examples of denoising noisy images using mean and median filters to remove different types of noise.
In this document
Powered by AI
Introduces presenter Haitham Abdel-atty Abdullah and supervisor Prof. Dr. Mostafa Gadal-Haqq.
Discusses signal reconstruction from noisy data, focusing on image de-noising as a vital pre-processing step.
Explains additive and multiplicative noise models that corrupt images, highlighting their distinct effects on the original signal.
Details types of noise including Gaussian, impulse, uniform, and periodic noise, each with specific causes and characteristics.
Identifies de-noising strategies under spatial and frequency domains, including linear and non-linear filters.
Examines linear filters designed to reduce noise, especially salt and pepper and Gaussian noise, with a focus on their strengths and weaknesses.
Demonstrates the mean filter technique for image smoothing with examples and its limitations in detail preservation.
Explains weighted mean filtering, emphasizing its advantages over simple averaging in preserving edges while reducing impulsive noise.
Describes frequency domain filtering methods using low-pass filters (LPF) and Fast Fourier Transform for noise reduction efficiency.
Concludes presentation with a Q&A session invitation and thanks the audience for their attention.
More Related Content
Image denoising
- 1. Presented By : HaithamAbdel-atty Abdullah Supervised By : Prof .Dr . Mostafa Gadal-Haqq
- 2. The processwith which we reconstruct a signal from a noisy one. Removing unwanted noise in order to restore the original image. Method of estimating the unknown signal from available noisy data”.
- 3. Remove noise Preserve useful information Image de-noising is an important pre-processing step for image analysis. Denoising
- 4. 𝑥𝑥 𝑛 𝑦 = 𝑥+ 𝑛 𝑛 = 𝑦 − 𝑥Noise
- 5. Additive Noise ModelMultiplicative Noise Model Noise signal gets added to the original signal to produce a corrupted noisy signal. Noise signal gets multiplied to the original signal to produce a corrupted noisy signal. The Additive noise model follows the following rule: The Multiplicative noise model follows the following rule:
- 6. Gaussian Noise Isevenly distributed over the signal This means that each pixel in the noisy image is the sum of the true pixel value and a random Gaussian distributed noise value.
- 7. Impulse Noise Theimpulse noise is of two types, Fixed value and random value. Fixed value (salt and pepper noise) It is generally caused due to errors in transmission. It has only two possible values For an 8-bit image, the typical value for pepper noise is 0 and for salt noise 255. The corrupted pixels are set alternatively to minimum or to maximum values giving the image Salt and Pepper Random value The random valued impulse noise can have any value between 0 and 255 hence its removal is very difficult.
- 8. Uniform Noise Theuniform noise cause by quantizing the pixels of image to a number of distinct levels is known as quantization noise. Uniform noise can be analytically described by :
- 9. Periodic Noise Source:electrical or electromechanical interference during image acquisition Easy to observe in frequency domain. Image with periodic noise frequency domain
- 10. De-noising Methods SPATIAL DOMAIN LINEARFILTERNON- LINEAR FILTER FREQUENCY DOMAIN Mean Filter Weighted Mean Filter Median Filter LPF on Fourier Transform
- 11. In spatialdomain , we deal with images as it is (simply we directly deal with the image matrix).
- 12. Linear filtersare used to remove certain type of noise. The linear filters work best with salt and pepper noise, and Gaussian noise. Mean filters. Simple to design . These filters also tend to blur the sharp edges . destroy the lines and other fine details of image .
- 13. One ofthe simplest spatial filtering operations. Replaces the centre value of the window with the average values of its all nearest pixels values together with itself. Advantage: Easy to implement Used to remove the impulse noise. Disadvantage: It does not preserve details of image. Some details are removes of image with using the mean filter.
- 14. 1/9 1/9 1/9 1/9 1/9 1/9 1/9 1/9 1/9 Origin x y Imagef (x, y) New value = 1/9*106 + 1/9*104 + 1/9*100 + 1/9*108 + 1/9*99 + 1/9*98 + 1/9*95 + 1/9*90 + 1/9*85 = 98.3333 Filter Simple 3*3 Neighbourhood 106 104 99 95 100 108 98 90 85 104 100 108 99 106 98 95 90 85 Original Image Pixels * The above is repeated for every pixel in the original image to generate the smoothed image
- 15. Extending border valuesoutside with values at boundary.
- 16. (a) Original Image (b)Image corrupted by Gaussian noise . (c)De noising by mean filter
- 17. Simple averaging(Mean filter) of neighboring pixels lead to over-smoothing Instead of weighting all neighboring pixels equally, assign higher weights to pixels that are closer to the pixel being convolved. Weighted Mean FilterMean Filter
- 18. Can preserveedges . Very effective at removing impulsive noise . They are more powerful than linear filters because they are able to reduce noise levels without blurring edges. Difficult to design than linear filters. Median Filter.
- 19. Median filteris one of the most important filters to remove random valued impulse noise. In this filter the value of corrupted pixel in noisy image is replaced by median value of corresponding window. The median is calculated by first sorting all the pixel values into ascending order and then replace the pixel being calculated with the middle pixel value Salt and pepper noise.
- 20. 99 Origin x y Imagef (x, y) Sort the pixels values 106 104 99 95 100 108 98 90 85 104 100 108 99 106 98 95 90 85 Original Image Pixels The above is repeated for every pixel in the original image. 9985 104100 108989590 106
- 21. 21 Original Salt & pepper %20 Denoisingby Median filter Original Salt & pepper %60 Denoising by Median filter
- 22. Frequency domainde-noising method is a kind of Transform Domain, filtering where low pass filters (LPF.) is used by using Fast Fourier Transform (FFT). The reason for doing the filtering in the frequency domain is generally because it is computationally faster to perform two 2D Fourier transforms and a filter multiply than to perform a convolution in the image (spatial) domain.
- 23. A low-passfilter attenuates high frequencies and retains low frequencies unchanged. Here de-noising is done by designing a cut-off frequency . Convolution in the Spatial domain is equivalent to multiplication in the frequency domain. Original Image Fourier Transform Apply LPF on FT Inverse Fourier Transform
- 24. Original Image withnoise Fourier Transform Inverse Fourier Transform
- 25.
- 26.