Sunteți pe pagina 1din 3

CSE3017 Computer Vision L T P J C

3 0 0 4 4
Pre-requisite: Nil
Version: 1.00
Objectives:
 To enable the students to understand the need and applications of computer vision
 To provide thorough understanding of Digital Image Processing which needed for
computer vision applications.
 To demonstrate and design a typical computer vision for real time problems.

Expected Outcome:
 Student will be able to
 Apply the feature extraction techniques for image description and recognition.
 Understand the pattern recognition approaches for implementing the visual system.
 Identify computer vision techniques in various real-time interdisciplinary projects.

Student Learning Outcomes 1,2,4,5,7,9


(SLO):

Module:1 IMAGE FORMATION AND 6 hours SLO:1


REPRESENTATION
Physical and biological aspects of Image acquisition – review of the digital camera – sampling
and quantization – Image quality – Color Camera - Color Balance

Module:2 MULTIPLE VIEWS AND IMAGE 7 hours SLO:1,4


ENHANCEMENT
Epipolar geometry- Trifocal geometry – More views – Point Operators: Pixel transform, Color
Transform, Histogram Equalization – Bandpass filters – 2D Convolution: Discrete &
continuous – Edge detection.
Module:3 FEATURE EXTRACTION 6 hours SLO:1,4
Introduction to lines and curves –corner detection - Hough Transform for lines & Curves –
Fitting Ellipses to image data - Classical Local Descriptor: SIFT & SURF- RANSAC

Module:4 IMAGE SEGMENTATION 6 Hours SLO:1,4


Active contours – Snakes – Dynamic snakes – Scissor – Level sets – Split and merge –
Watershed – Graph based segmentation – Graph cuts and energy based methods – Application:
Medical image segmentation

Module:5 MOVEMENT ANALYSIS 6 hours SLO:4,7


Importance of visual motion – The problems of motion analysis – The motion field of Rigid
objects – Optical flow: Horn & Schunck, Lucas & Kanade, Pyramids: Laplacian, Gaussian –
Estimating the motion field – 3D Motion and structure from a sparse & Dense motion field -
Kalman filter for object tracking
Module:6 RECOGNITION, CLASSIFICATION & 9 hours SLO:4,9
CLUSTERING
Patterns and pattern classes – template matching – Introduction to classification – Decision
theoretic methods – structural and syntactic classifiers – Clustering techniques – similarity
measures – hierarchical methods – K – Means algorithm – Cluster evaluation methods.
Convolution neural networks, Region-based CNN, fully convolution networks, Multi-modal
networks, Hybrid learning methods.

Module:7 COMPUTER VISION APPLICATIONS 3 hours SLO:5,9


Face recognition application: personal photo collections – Instance recognition application:
Location recognition – Machine learning applications : Deep voting and structured regression for
microscopy image analysis. Chest radiography pathology categorization via transfer learning

Module:8 CONTEMPORARY ISSUES (To be 2 hours SLO:2


handled by experts from industry)
Total Lecture: 45 hours
Text Books:
1 Richard Szeliski, ”Computer vision: Algorithm and Applications”, Springer- verlag, London,
2010
2 Sridhar, Digital Image processing, Oxford University Press , 2011
Reference Books:
1 David A. Forsyth and Jean Ponce, “Computer Vision: A Modern Approach”, Prentice
Hall, Pearson Education, 2nd Edition, 2012
2 Rafael C. Gonzalez, Richard E.Woods & Steven L. Eddins, “Digital Image Processing
using MATLAB”, Pearson Education 2nd Edition, 2009.
3 E.Trucco, A.Verri, “Introductory techniques for 3D Computer vision”, Prentice Hall,
1998
4 Bin Fan, Zhenhua Want, Fuchao Wu, “Local Image Descriptor: Modern approaches”,
Springer, 2015
5 Rafael C. Gonzalez & Richard E.Woods, “Digital Image Processing”, Pearson Education
3rd Edition, 2009.
6 Image Processing Research Group “www.iprg.co.in”
7 The computer vision home page “https://www.cs.cmu.edu/~cil/vision.html”
8 Deep Learning for Medical Image Analysis, S. Kevin Zhou, Hayit Greenspan, Dinggang
Shen, Academic Press, 2017, ISBN 0128104090, 9780128104095
SLO: 6,13
Sample Projects:
1. Tracking of object based on color feature.
2. Tracking of object based on texture feature.
3. Image segmentation using feature set.
4. Detection of multiple objects in the image and naming them.
5. Detection of pedestrian on road.
6. Listing no. of vehicles in parking yard with their license no.
7. Motion tracking
8. Motion segmentation
9. Image classification using using Deep learning techniques

S-ar putea să vă placă și