AI361 UPM Othman Soufan Week 14 Fall 2024 Image Processing PDF

Image Processing Week 14&15: Color Image Processing & Other Topics Dr. Othman Soufan ([email protected]) Dr. Karim Said Barsim ([email protected]) UPM University Fall 2024 What you learned last week/ what sticks...

Image Processing Week 14&15: Color Image Processing & Other Topics Dr. Othman Soufan ([email protected]) Dr. Karim Said Barsim ([email protected]) UPM University Fall 2024 What you learned last week/ what sticks out the most? Some Reminder! Final project presentations – Week 16 See Appendix for the template Weight: 6% 10 mins + 5 mins. Q&A 3 What you learned last week/ what sticks out the most? Week’s Outline Color Fundamentals Color models 5 Bird’s Eye on the Course Visual Representation Storage Fourier Transform Image 𝑓 (𝑥 , 𝑦 ) 𝑓 (𝑥 , 𝑦 , 𝑡) Wavelet 𝑝 ( 𝑧𝑘 ) ∼ 𝒩 ( 𝜇 , 𝜎 Linear Transform 2 ) Algebra Statistical Probabilistic Treatment Deep Treatment Learning Numeric Analytic Representation Representation 6 Why use color in image processing? Applications of Color Image Processing 8 Applications of Image Processing Harsh et al., 2020. 9 Image from: https://www.sciencedirect.com/science/article/pii/S0960077920305865 Any other applications? Excited! Let’s do some programming! Retina – The sensor chip 12 Color Fundamentals Human eye can decipher thousands of color shades and intensities, compared to only a few dozen shades of grey color 13 How do we recognize colors? 14 Absorption of Light 15 Color Models 16 Primary Colors 17 RGB Color model 18 CMY Color model 19 20 Maturity of color Three elements in color: "Hue" "Lightness" "Saturation". We call these maturity of color (three elements of color). "Hue" is to the color properties such as red, blue, and green "lightness" is to the brightness of the color, "saturation" is to the vividness of the color. Hue 21 Saturation "Saturation" refers to the intensity of color and the degree of vibrancy of color. Saturation has a strong and a bright color is high, weak and the dull color is low. 22 Brightness "Lightness" is the degree of "brightness" of color. The color gets brighter as the lightness gets higher, and it gets darker when it gets lower. 23 "Achromatic" vs "chromatic" 24 Trichromatic Coefficients 25 Trichromatic Coefficients - Example 26 CIE Chromaticity Diagram 27 RGB color cube 29 RGB Example 30 CMY & CMYK Color Space 31 HSI Model 32 HSI Model 33 From RGB to HSI 34 Why convert? - Images can not be manipulated under the RGB model, we need to convert them to the HIS model in order to do so. - Steps: a. Convert to HIS model b. Perform image processing techniques under HIS. c. Finally convert back to RGB. 35 From RGB to HSI: Intensity Calculation 36 From RGB to HSI: Saturation Calculation 37 From RGB to HSI: Chroma Calculation 38 From RGB to HSI: Hue Calculation 39 Example: Convert to HSI (R = 100, G = 150, B = 200) Color Processing 41 Pseudo-color image processing 42 Pseudo-color image processing - Intensity slicing 43 Intensity slicing (cont.) 44 Intensity slicing (cont.) 45 Application 46 Intensity slicing is powerful when exact gray level is known Application 47 Image Segmentation 48 Understanding Image Segmentation Definition: Image Segmentation is the process of partitioning a digital image into multiple segments (sets of pixels) to simplify or change the representation of an image into something more meaningful and easier to analyze (attributes). 49 Applications: Autonomous Vehicles 50 Applications: Agriculture 51 Types of Segmentation 1. Edge-Based Segmentation 2. Threshold-Based Segmentation 3. Region-Based Segmentation 4. Cluster-Based Segmentation 5. Watershed Segmentation 52 Fundamentals of Image Segmentation 53 Excited! Let’s do some programming! Image Segmentation - Gradient-based Approaches (e.g., Edge- based) 55 Revisit 1st Derivative & 2nd Derivative 56 1st Derivative & 2nd Derivative 57  Revisit 1st Derivative & 2nd Derivative Taylor Series Taylor Series (change in x → measured in pixel units → +1 for sample preceding x, forward difference) Taylor Series (change in x → measured in pixel units → -1 for sample following x, backward difference) 58 Intensity Differences using 1st Derivative forward backward difference difference Central difference 59 Intensity Differences using 1st Derivative forward backward difference difference Central difference (1st order derivative) Central difference (2nd 60 Central Derivatives 61 62 Point Detection Laplacian 63 Point Detection 64 Example 66 Line Detection 67 Line Detection 68 Edge Detection 69 Step-edge Ramp-edge Roof-edge 70 Edge Detection 71 Edge Detection 72 Colorization in digital image processing transforms grayscale images into color, using algorithms to interpret intensity and texture, adding depth and vibrancy to monochromatic visuals. 73 Thanks! Let me answer your questions... [email protected] 74