Weekly Topics for CSE190 (wi02)

Week 1:
 [Monday]
 course overview, policies
 overview of textbook website: images, supplementary material
 discussion of Matlab
 smoothing

 [Wednesday]
 Linear systems
 superposition
 convolution and its properties
 1st difference kernels

 [Friday]
 2D convolution
 Gaussian
 binomial kernel
 separable filtering
 sampling and aliasing

Week 2:
 [Monday]
 smoothed gradient
 directional derivative
 Canny edge detection
 nonmaximal suppression
 hysteresis thresholding

 [Wednesday]
 Laplacian of Gaussian
 zero crossings
 frequency domain
 Fourier transform background
 orthogonal basis
 DFT, IDFT

 [Friday]
 DFT: 2D case
 magnitude and phase
 example DFTs: box, cosine
 DC component
 convolution theorem

Week 3:
 [Monday]
 (holiday)

 [Wednesday]
 properties of the DFT
 periodic convolution
 zero padding
 convolution matrix
 DFT matrix
 frequency domain filtering

 [Friday]
 ideal LPF (lowpass filter)
 Butterworth LPF
 Gaussian LPF
 ringing
 Laplacian of Gaussian in frequency domain
 FT of differentiation
 Derivative/Difference of Gaussian filter
 Gabor filter

Week 4:
 [Monday]
 brief discussion of FFT
 physics of color
 cone absorption spectra
 color spaces: RGB, HSI 

 [Wednesday]
 theories of color vision
 trichromacy
 opponent channels
 dual process theory
 color distributions/histograms
 chi squared test

 [Friday]
 histogram equalization
 median filter

Week 5:
 [Monday]
 overview of image compression
 entropy
 lossless encoding
 Huffman coding
 arithmetic coding

 [Wednesday]
 lossy encoding
 transform coding
 DCT basis, JPEG

 [Friday]
 midterm review

Week 6:
 [Monday]
 (midterm)

 [Wednesday]
 overview of image analysis
 binary image processing
 connected components labeling
 dilation, erosion, etc.
 centroid
 scatter matrix
 
 [Friday]
 eigenvalues and eigenvectors
 principal axes
 diagonalization of the second moment matrix
 rotation matrix
 alignment using 1st and 2nd moments
 overview of planar transformation models using point correspondences

Week 7:
 [Monday]
 (holiday)

 [Wednesday]
 least squares solution of affine transform
 pseudoinverse
 corner detection
 windowed-image second moment matrix
 neighborhoods of rank 0, 1, and 2

 [Friday]
 line detection: local vs. global
 Hough transform
 overview of geometric hashing

Week 8:
 [Monday]
 Principal Components Analysis (PCA)
 feature vectors
 capturing the most variance
 eigenvectors of the covariance matrix
 centering the data
 truncating small principal components
 sketch of eigenfaces

 [Wednesday]
 (short class due to CAPE teaching survey)
 more information on eigenfaces
 reconstruction from principal components
 small-covariance matrix trick

 [Friday]
 overview of image segmentation problem
 motivation based on human-labeled segmentations
 central vs. pairwise grouping
 k-means clustering algorithm
 local minima

Week 9:
 [Monday]
 measurement of image motion: overview
 block matching
 aperture problem
 brightness constancy equation
 optical flow: the 1D case
 
 [Wednesday]
 optical flow: the 2D case
 1 equation with 2 unknowns
 smoothness assumption
 Lucas-Kanade algorithm
 connection to windowed image second moment matrix
 validity of 1st order Taylor series

 [Friday]
 prototypes
 decision boundary
 minimum-distance classifier
 normalized correlation coefficient

Week 10:
 [Monday]
 risk/loss function
 posterior, prior, and class conditional probability
 Bayes classifier
 Mahalanobis distance

 [Wednesday]
 Final lecture: overview of biometrics
 false accept, false reject
 ROC curve 
 fingerprint recognition
 iris recognition

 [Friday]
 (final review)