Course Archives Statistical Quality Control & Operations Research Unit
Course: Pattern Recognition
Level: Postgraduate
Time: Currently not offered
Past Exams

Syllabus: Elements of Image processing and Analysis: Biology and physics of image formation and recognition-Digital images. Components of image processing system sensors, digitizer, processers, display unit, and hard copier. Mathematical preliminaries required vector algebra, orthogonal transformations, probability and statistics, fuzzy sets and properties, and mathematical morphology. Image processing Greyvalue histograms, Greyvalue distributions & statistics, thresholding & segmentation; Point operations Histogram transforms, pixels, gridding & quantization; Patterns and classes; image enhancement, image smoothening, image sharpening, image restoration, image compression, and image registration. Image analysis image segmentation, edge and line detection, feature extraction, and description. Recognition deterministic approaches, statistical approaches, fuzzy mathematical approach, syntactic approach, and morphological approaches.
Statistical and Fuzzy Mathematical Approach Pattern Recognition: Bayesian decision theory; Maximum likelihood and parameter estimation; Nonparametric techniques; Qualifying structure in pattern description and recognition; Grammar based approach; Neural pattern recognition.
Structural and Syntactic Pattern Recognition:
Segmentation: Detection of discontinuities Pont, Line, Edge and Combined detection; Edge linking and boundary detection Local processing, Global processing via Hough transform and Graph Theoretic techniques; Thresholding Foundation, The role of illumination, Simple Global Thresholding, Optimal Thersholding, Thershold selection based on boundary characteristics, thresholds based on several variables; Region-Oriented segmentation Basic formulation, region growing by pixel aggregation, region splitting and merging, morphologic segmentation, watersheds; The use of Motion in Segmentation spatial and frequency domain techniques; Texture segmentation pattern spectra and Granulometries.
Boundary and Region Representation and Description: Representation and Description: Representation schemes chain codes, polygonal approximations, signatures, boundary segments, the skeleton of a region; Boundary descriptors Simple descriptors, shape numbers, Fourier descriptors, moments; Regional descriptors simple descriptors, topological descriptors, texture, moments; Morphology Dilation, erosion, opening, closing, Hit or Mist Transform, basic morphological algorithms, extensions to grayscale images; Relational Descriptors. Recognition and Interpretation: Decision Theoretic Methods Matching (Minimum distance classifier, Matching by correlation); Optimum statistical classifier (Foundation, Bayes classifier for Gaussian pattern classes); Neural networks (Background, perceptron for two pattern classes, training algorithms, multiplayer feedforward neural networks). Structural methods: match shape numbers, string matching, syntactic methods. Interpretation: Object measurements Size, shape and orientation: Statistics of size distributions, resolution and scale; shape analysis, orientational statistics; Stereological models and microstructural analysis; Analysis of 3D data sets

Reference Texts:
1. Pattern Classification: By Duda RO, PE Hart and DG Stork, John Wiley & Sons, NY, 2nd Ed, 2001.
2. Structural Pattern Recognition: By T. Pavlidis, Springer Verlag, NY, 1977.
3. Image analysis and Mathematical Morphology: By J. Serra, Academic Press, 1982.
4. Digital Image Processing: By RC Gonzalez and RE Woods, Addison Wesley Publishing Company, 1992.
5. Digital Image Processing and Analysis: By B. Chanda and D. Dutta Majumdar, Prentice Hall, India.
6. Fuzzy Mathematical Approaches to Pattern Recognition: By S. K. Pal, and D. Dutta Majumdar, Wiley.

Top of the page

Past Exams
15.pdf 16.pdf 17.pdf
15.pdf 16.pdf 17.pdf
Supplementary and Back Paper

Top of the page

[Indian Statistical Institute]