Image Quantization

This slides guides you through the basic of image quantization used during the process of image generation.

1. Image Quantization Digital Image Generation Anish Chhetri 073-BGE-504

2. Key stages in digital image Generation Image captured by sensor (camera) are in continuous voltage waveform Continuous in term of x and y coordinates and amplitude Digital image are represented in digital form i.e. discrete signals conversion of captured continuous signal into discrete signal Sampling Quantization

3. Image Quantization Process of digitizing the amplitude value of the continuous signal Continuous grey level intensity is converted in discrete form Depicts the grey level resolution of image

4. General Steps in Image Quantization Measuring the grey level intensity of the signal in fixed interval in time Value obtained in each instant of time is converted in number and stored This number depicts brightness value of a particular point Such point is called pixel QUANTIZATION

5. Image Matrix Represents the intensity value or pixel value For n bit image, intensity value ranges form 0 – 2 n-1

6. Drawbacks of quantization Generally irreversible Results in loss of information Introduces distortion which cannot be eliminated

7. Quantizing a grey-level image

8. Quantizer Used for quantization Amount of distortion depends upon the quantizer Good quantizer results in better quantization of image

9. Classification of Quantizer Quantizer Uniform Quantizer Non-uniform quantizer Zero Memory Quantizer

10. Zero Memory Quantizer Simplest type of quantizer Quantizing a sample is independent of other sample Maps amplitude variable to a discrete set of quantization levels, {r 1 ,r 2 …, r l } Based on simple comparison / thresholding with certain values, t k t k = transition/ decision level r l = reconstruction level

11. Zero Memory Quantizer

12. Uniform Quantizer Simplest form of zero memory quantizer Quantization level are uniformly spaced Shows absolute change in amplitude of stimulus t k and r k are equally spaced Mathematically given as:

13. Uniform Quantizer

14. Non-Uniform Quantizer Quantization levels are not necessarily equally spaced Logarithmic relation between quantization levels Shows proportional change in amplitude of stimulus Better for human perception Quantization level are assigned from histogram analysis

15. Non-Uniform quantization, 4 level Uniform quantization, 4 level

16. Questions? Comments?