BSCS 621 - Paper Pattern and Course Material

BSCS 621 - Paper Pattern and Course Material
Paper Pattern
Section A: Answer the following Questions. Use proper vocabulary, Mathematical Expressions,
Illustrations, Pseudo-code and Examples to get full credit.
Section B: Problem Solving and Derivations
Section C: Visualization and Lab Practice
Course Material
Below are the contents that we discussed and covered in class. To clarify the topic numerical examples
from each topic has been done.
Topic
Lecture Discussion
Introduction 1) What a vision system does?
2) What a CV System is supposed to do?
3) Image Processing Pipeline.
4) Related Disciplines and difference among them.
5) Handout on Relationship between Pixels uploaded.
Gonzales Reference Reading
1.1- What is DIP?
1.4- Fundamental steps in DIP.
2.4.2 -Representing Digital Images.
2.5- Some Basic Relationship between
Pixels.
Filtering
1) What is filter and what are various uses of filters?
2) Image Space or Spatial Space Concept.
3) Neighborhood Concept4- How to apply filter on
image?
4) Derivation of Box Filter.
5) Matlab usage and Code Provided for Practice and
Understanding.
Edge
Detection
1) What is Image Segmentation?
2) Various Image Segmentation Approaches.
3) What are edges? What are the possible causes of
Edges?
4) Fundamental Edge Detection Techniques (1st and
2nd Derivative, Marr-Hildreth Edge
Detector, Canny Edge Detector)
5) Derive Sober, Robert and Prewit.
6) Block Diagram and detailed Explanation of Canny
Edge Detector.
7) How to visualize 1st derivative, 2nd derivative and
Filters is discussed.
8) Derivation of LOG, Approximation of LOG as DOG
(Difference of Gaussian).
9) Why separate/decompose filters along x and y? OR
Why prefer 1D convolution over 2D
convolution?
10) Worksheet done in class on Filtering and Edge
Detection
11) Matlab usage told in class and Code is also
provided for Practice and understanding.
3.1-Background
3.4 - Fundamentals of Spatial
Filtering
3.5- Smoothing Spatial Filters
(Exclude non-linear filters)
3.6-Sharpening Spatial Filters
(Exclude Highboost Filtering)
10.1- Fundamentals of Image
Segmentation.
10.2 - Point, Line and Edge Detection
(Complete, Exclude only Regional
Processing)
Hough
Transform
1)
2)
3)
4)
5)
6)
7)
Template
Matching
1)
2)
3)
4)
5)
6)
7)
Histogram
Computation,
Histogram
Smoothing,
Histogram
Equalization
Histogram
Thresholding
Histogram
of Oriented
Gradients
(HOG)
1)
2)
3)
4)
5)
6)
7)
1)
2)
3)
4)
5)
Slope -Intercept Line Representation
Polar Representation of Line
Concept of Parameter/Hough Space
Hough Transform Algorithm Discussion
Numerical Exercises to understand input , out put
and voting concept of Hough Transform.
Visualizing and interpreting Hough space.
Matlab Usage discussed in class and Code is also
provided for Practice and understanding.
Correlation vs. Convolution
Basic Template Matching Steps
Different Ways to Normalize. Why NCC?
Convolution as Dot product between two vectors.
Different Ways to slide filter on image ( check
matlab options 'full', 'same' and 'valid')
Student are asked to verify their numerical on
matlab.
Code provided for Practice and Understanding.
10.2 - Global Processing
3.4 - Fundamentals of Spatial
Filtering
12.2.1 - Matching by correlation
What is Histogram? How to compute it?
How Histogram is turned into pdf?
How to change image brightness and image
contrast using histogram processing.
What information can be extracted from
histogram visualization?
Finding Thresholds that minimizes error.
Noisy Histogram vs. Smooth Histogram
Numerical done and matlab code is provided.
3.3. Histogram Processing, Histogram
Equalization ( Derivation excluded)
Concept of Feature Space and Feature Vector
Block Diagram of HOG Algorithm
Explanation of each step with matlab code.
Numerical on computation of cell and block
histogram.
Matlab Code Provided for Practice and
understanding.
Original Paper :
dalal_triggs_cvpr2005
10.3.1 The basics of intensity
thresholding
10.3.2 Basic Global Thresholding
https://chrisjmccormick.wordpress.c
om/2013/05/09/hog-persondetector-tutorial/