This document provides an overview of color image processing. It discusses pseudo color image processing which involves assigning different colors to intensity values in a grayscale image. Full color image processing techniques are also described, including color transformation, intensity modification, color complements, color slicing, tone corrections, and color compression. Color models like RGB, CMY, and HSI are introduced. Various color image processing operations and their applications are explained such as color conversion, intensity modification, tone corrections, sharpening, and smoothing.
Pests of mustard_Identification_Management_Dr.UPR.pdf
Color Image Processing
1. INDIAN INSTITUTE OF TECHNOLOGY ROORKEE
Colour Image Processing
Presented By
Neelisetty Sesha Sai Baba
22555010
MTech 1st Year
Packaging Technology
Course: Printing Technology
Course Code: PPN 545
Presented To
Vibhore Kumar Rastogi
Assistant Professor
Department of Paper Technology
2. 2
Contents
• Why CIP
• CIP
• Types
• Basics of Colour Science
• Colour Models
• Pseudo Colour Image Processing (PCIP)
• Colour Transformation in PCIP
• Full Colour Image Processing (FCIP)
• Colour Transformation in FCIP
• Intensity Modification
• Colour Complements
• Tone Corrections
• Smoothing and Sharpening
• Colour Compression
3. 3
WHY Colour Image Processing
• Colour is very good Descriptor
• Easy to Extract Objects of Interest
• Can distinguish easily in colour images (Less variations in B/W)
4. 4
Colour Image Processing
• Extracting useful information out of an image by processing it
• Applications include
– Colour Conversion
– Colour Transformation
– Intensity Modification
– Colour Complements
– Colour Slicing
– Tone Correction
– Smoothing and Sharpening
5. 5
Types
• Pseudo Colour Image Processing
– Processing of B/W image
– Intensities are close in B/W image so difficult to distinguish so we use PCIP to differentiate
regions based on intensity
• Full Colour Image Processing
– To identify or focus on a particular area or colour space we use FCIP
– Applications in Colour Slicing, Colour Complements, Tone Corrections, Smoothing and
Sharpening
6. 6
Colour Science
• Light is basically 2 Types
– Achromatic – No colour component, Defined by Intensity
– Chromatic – With Colour Component, Defined by Wavelength
• Quantification of Light
– Radiance – Total Energy emitted by light source (Watts)
– Luminance – Total Energy perceived by humans (Lumens)
– Brightness – Subjective Measurement
• Utility
– Machine Quantification – By use of colour models like RGB, CMY(K)
– Human Quantification – By use of visible characteristics like Hue Saturation and Intensity
7. 7
Colour Models
• RGB Model for monitors and light
– Red – 65% cones are red sensitive. CIE Standard Wavelength 700 nm, cone perception 450
to 700 nm
– Green – 33% cones are green sensitive. CIE Standard Wavelength 546.1 nm, cone
perception 410 to 650 nm
– Blue – 2% cones are blue sensitive. CIE Standard Wavelength 435.8 nm, cone perception 400
to 550 nm
• CMY Model for printers
– Cyan is combination of blue and green
– Yellow is combination of green and red
– Magenta is combination of red and blue
𝐶 = 1 − 𝑅
𝑀 = 1 − 𝐺
𝑌 = 1 − 𝐵
8. 8
Cont.
• HSI Model
– Hue – Dominant Wavelength in a mixture of
colours
– Saturation – Purity of the colour.
– Intensity – Chromatic notion or brightness of
the colour
Red Pink
Spectrum Colours (Red) – Fully Saturated
Other Colours (Pink) = Red + White –
Unsaturated
• Hue and Saturation defines chromaticity
or function of colour while brightness is
the function of intensity
9. 9
Pseudo Colour Image Processing
• We can only interpret few range of grey scale intensity
• So we assign different colours at different intensities to interpret the data easily
• Called False Colour Image ProcessingNo real colours
• Assigning of different colours at different intensity values of grey level
• Done by use of intensity slices
– 2D image to 3D image conversion with Intensity as third coordinate
– Slicing based on intensity
– Assigning colours to each slit
– Intensity Ranges from 0 (Black) to 𝐿−1(White)
– P no of planes divides intensity with P+1 intervals
– Each plane is assigned a different intensity of colours RGB
11. 11
Colour Transformation of PCIP
Green Transformation
Blue Transformation
Red Transformation
F(x,y)
𝐹𝑅(x,y) = a F(x,y)
𝐹𝐵(x,y) = c F(x,y)
𝐹𝐺(x,y) = b F(x,y)
𝑆 = 𝑇 ∗ 𝑟
S – Transformed Image Intensity
T – Transformation Factor
r – Original Image Intensity
12. 12
Examples
𝐹𝑅(x,y) = K F(x,y)
𝐹𝐺(x,y) = L F(x,y)
𝐹𝐵(x,y) = M F(x,y)
K is close to 0 while L and
M are close to 1, L > M
13. 13
Cont.
𝐹𝑅(x,y) = F(x,y)
𝐹𝐺(x,y) = 0.33 F(x,y)
𝐹𝐵(x,y) = 0.11 F(x,y)
Red is more dominant throughout
The Intensity followed by Green and
Blue
Before
After
15. 15
Full Colour Image Processing
• Use of Processing Techniques to Full Colour Image
• Used for image intensity modification, Colour Complementing, Colour Slicing,
Tone Corrections, Sharpening and Smoothing operations
• The focus is on particular colour or intensity of colours
• Mainly of two types
– Per Colour Plane Processing where each colour is processed
separately and then combined to give full colour processed image
– Vector Processing where the colour is considered to be vector in
colour space as component of RGB and then processed as whole
16. 16
Colour Transformation in FCIP
• We can use any processing with any colour space
• But we have to look after convenience and cost
associated with conversion
𝑆𝑖 = 𝑇𝑖 ∗ (𝑟1, 𝑟2, 𝑟3 )
𝑆𝑖 – Transformed Image Intensity
𝑇𝑖 – Transformation Factor
𝑟1, 𝑟2, 𝑟3 – Original Image Intensity
17. 17
Intensity Modification
• Done to alter the intensity of coloured planes in an image
• To improve the detailing in the image
• RGB colour space
𝑆𝑖 = 𝐾 ∗ (𝑟𝑖 )
• CMY colour space
𝑆𝑖 = 𝐾 ∗ (𝑟𝑖 ) + (1−K)
• HSI colour space
𝑆3 = 𝐾 ∗ (𝑟3 )
𝑆1 = (𝑟1 )
𝑆2 = (𝑟2 )
• HSI has got least processing
• But conversion costs associated with converting image
from RGB colour space to HSI colour space is to be kept in
mind
18. 18
Colour Complements
• Every colour has it’s own complement
• So to create complement image of a given colour
image we use colour complementing
• It is same as creating negative of a Black and White
Image
𝑆𝑖 = 𝐿 − 1 − (𝑟𝑖 )
19. 19
Colour Slicing
• Done to extract details of a particular colour
• Rest of the colours will be removed so called colour
slicing
𝑆𝑖 =
0.5 𝑟𝑖 𝑖𝑓 𝑟
𝑗 − 𝑎𝑗 > 𝑊/2
𝑟𝑖 𝑂𝑡ℎ𝑒𝑟𝑤𝑖𝑠𝑒
• We will compare the intensity of the colour with our
colour of interest. If matches, we will leave it as it is
else it’s intensity is reduced
to half.
20. 20
Tone Correction
• Images may have flat tone, light tone or
dark tone.
• We will apply the tonal corrections to
image to make the appearance of the
image better
– Flat Tone to Bright Tones
– Light Tone to Dark Tone
– Dark Tone to Light Tone
21. 21
Smoothing and Sharpening
• To improve the transition from surface
to surface we use colour smoothing
C(x,y) =
1
𝑘
∗ 𝛴 R(x,y)
• To improve the details in the image we
use colour sharpening
C(x,y) = 𝑘 ∗ 𝛴 R(x,y)
22. 22
Colour Compression
• Compression is the process of reducing
or eliminating redundant and/or
irrelevant information
• In case if in a compressed image 1 bit
of data represents 230 bits of data in
the original image
• Hence compressed image could be
transmitted over internet in seconds as
compared to original image which will
take more than that
23. 23
Effects of Improper Colour Compression
• Phones can compress any image from
Different colour spaces to RGB colour
space by use of Google Colour
Encoding Engine.
• If the colour is beyond illuminance of
255, then it is not able to process it
making devices to hang
24. 24
• This sums upto 254 which is not a
problem
• But Google Colour Engine rounds to
next integer making it 256 which is
beyond 255
• So devices hang as they can’t able to
process this image.
25. 25
Reference
• https://youtu.be/svgZodJgKaU
• https://youtu.be/IbHPLbng_d4
• https://youtu.be/Q25579dA-YY
• Azad MM, Hasan MM, Mohammed Naseer K (2017) Color image processing in digital
image. Int J New Technol Res (IJNTR) 3(3): 56–62. ISSN: 2454–4116
• https://slideplayer.com/slide/6399279/
• https://slideplayer.com/slide/13671016/