IF4092 Computer Vision Syllabus:
IF4092 Computer Vision Syllabus – Anna University PG Syllabus Regulation 2021
COURSE OBJECTIVES:
Articulate & apply standard computer vision concepts
Implement standard image processing tasks
Applying Clustering concept for Image Classification
Identify practical constraints in computer vision application
Architecture of an existing computer vision pipeline based on deep learning models
UNIT I COMPUTER VISION
About Computer Vision. Components of an Image Processing System. Image Resolution. Image Formats. Colour Spaces. Fundamental of Image Processing. Visual Inspection System. Biomedical Imaging Methods. Image Thresholding. Based Image Retrieval. Human Visual Inception. Image Formation. Geometric Properties. 3D Imaging. Stereo Images.
UNIT II PIXEL-BASED MANIPULATIONS & TRANSFORMATION
Visual properties. Pixel colour manipulation. Colour Change with Pixel Position. Colour Change with Pixel Distance. Colour Change with Trigonometric Functions. Randomness. Drawing with existing images. Blending multiple images. Image transformation. Image orientation. Image resizing. Affine transform. Known Affine Transformations. Unknown Affine Transformations. Perspective transform. Linear vs. polar coordinates. Three-dimensional space. General pixel mapping.
UNIT III STRUCTURE IDENTIFICATION
Image preparation. Conversion to grayscale. Conversion to a black-and-white image. Morphological operations (erode, dilate). Blur operations (smoothing)Edge detection. First Derivative Edge Detectors. Second Derivative Edge Detectors. Multispectral Edge Detection. Line detection. Circle detection. Contours processing. Finding the contours. Bounding box. Minimum area rectangle. Convex hull. Polygon approximation. Testing a point in contour. Checking intersection. Shape detection. Moravec Corner Detection. Harris Corner Detection. FAST Corner Detection. SIFT.
UNIT IV CLUSTERING IMAGES & IMAGE RETRIEVAL
About Transfer Learning. Extract features. SciPy Clustering Package. K-Means Clustering. Clustering Images. Principal Components. Clustering Pixels. Hierarchical Clustering. Spectral Clustering. Fast Fourier Transforms. -Based Image Retrieval. Indexing Images. Searching the Database for Images. Querying with an Image. Benchmarking and Plotting the Results. Ranking Results Using Geometry.
UNIT V IMAGE CLASSIFICATION USING DEEP LEARNING
Working with Image Datasets. k-NN: A Simple Classifier. k-NN Hyperparameters. Gradient Descent. Loss Functions. Stochastic Gradient Descent (SGD). Regularisation. The Perceptron Algorithm. Backpropagation and Multi-layer Networks. Weight Initialization. Constant Initialization. Uniform and Normal Distributions. CNN Building Blocks. Image Classification.
SUGGESTED ACTIVITIES:
1: Identify and List various noises in the Image.
2: Identify Image Manipulation
3: Add colour descriptors and improve the search results.
4: Hierarchical k-means is a clustering method that applies k-means recursively to the clusters to create a tree of incrementally refined clusters
5: Image Classification using CNN
TOTAL:45 PERIODS
COURSE OUTCOMES:
CO1: Understand the basic knowledge, theories and methods of computer vision.
CO2: to understand the essentials of image processing concepts through mathematical interpretation.
CO3: Demonstrate a knowledge of a broad range of fundamental image processing and image analysis techniques
CO4: Apply Clustering algorithms for clustering.
CO5: Analyse cognitive tasks including image classification, recognition and detection through deep learning.
REFERENCES
1. Pro Processing for Images and Computer Vision with OpenCV, Bryan WC Chung, Apress, 2017
2. Programming Computer Vision with Python, Jan Erik Solem, O’Reilly Media, 2012
3. A PRACTICAL INTRODUCTION TO COMPUTER VISION WITH OPENCV, Kenneth Dawson-Howe, Wiley, 2014
4. Practical Computer Vision Applications Using Deep Learning with CNNs: With Detailed Examples in Python Using TensorFlow and Kivy, Ahmed Fawzy Gad, Apress. 2018
5. Computer Vision Principles, Algorithms, Applications, Learning E.R. Davies, Academic Press, 5th edition, 2017