CSE 5524 Topic Summaries (by slide)
Introduction - ANS-
*Image Formation* - ANS-*Imagers (CCD, CMOS)*
*Cable Quality*
*Images*
- size, interlacing, storage
*Color to gray conversion*
*MATLAB functions*
*Noise Removal* - ANS-*Noise removal for images*
- At local neighborhood
*Median filter*
- Good for salt-and-pepper noise
- Preserves edges
- Not separable
*Average filter*
- smooth image
- separable filter
*Gaussian filter*
- Weight influence of pixels by their distance to their center pixel
- Normal distribution
- Spread parameters
- Separable filter**
*Edge Detection* - ANS-*Interesting things happen at an edge*
- Object boundaries
*Look for derivatives/gradients in image*
- First and second derivatives/gradients
*Classic gradient operators*
- Sobel, Prewitt
- Gaussian derivatives
*Primary problem in edge detection is dealing with image noise*
- Smoothing and hysteresis thresholding
**Canny edge detector** (IMPORTANT!)
*Image Pyramids* - ANS-*Image pyramids as multi-resolution image representation*
- Gaussian pyramid
- Laplacian pyramid
*Two fundamental operations for pyramids*
- First operation blurs and samples the input
, - Second operation *interpolates* the blurred and sampled image to estimate the original
*Laplacian Error Pyramid*
- Error is difference between interpolated estimate and original
- Original signal can be recovered exactly by interpolating, then summing all the levels of the
error pyramid
- Can be represented very efficiently
*Useful for image coding/compression and progressive transmission*
*Gaussian pyramid useful for other tasks*
*Region Segmentation* - ANS-*Region segmentation*
*Background subtraction*
- Simple differences
- Statistical distance
- Multi-modal
*Morphology*
- Dilation to enlarge region
- Erosion to reduce region
- Closing (dilation + erosion)
- Opening (erosion + dilation)
*Region Growing*
- Connected components
- sequential
- recursive
*2D Shape* - ANS-*Given a 2-D binary shape, use representations/properties to characterize
the region*
- Recognition or matching
*Methods*
- Chain code
- Quadtree
- Medial axis
- Bounding box, extremal points
- Area, centroid
- Perimenter, compactness, circularity
- Signatures
- Moments
Hough Transform
- Used to identify/represent shapes in images by searching all possible parameters
*PCA and Face Recognition* - ANS-*Reduce high-dimensional input to lowerdimensional
"sub-space"*
- Beneficial for recognition
*PCA offers linear approximation to the sub-space
which can be reduced to only the major sub-space
Introduction - ANS-
*Image Formation* - ANS-*Imagers (CCD, CMOS)*
*Cable Quality*
*Images*
- size, interlacing, storage
*Color to gray conversion*
*MATLAB functions*
*Noise Removal* - ANS-*Noise removal for images*
- At local neighborhood
*Median filter*
- Good for salt-and-pepper noise
- Preserves edges
- Not separable
*Average filter*
- smooth image
- separable filter
*Gaussian filter*
- Weight influence of pixels by their distance to their center pixel
- Normal distribution
- Spread parameters
- Separable filter**
*Edge Detection* - ANS-*Interesting things happen at an edge*
- Object boundaries
*Look for derivatives/gradients in image*
- First and second derivatives/gradients
*Classic gradient operators*
- Sobel, Prewitt
- Gaussian derivatives
*Primary problem in edge detection is dealing with image noise*
- Smoothing and hysteresis thresholding
**Canny edge detector** (IMPORTANT!)
*Image Pyramids* - ANS-*Image pyramids as multi-resolution image representation*
- Gaussian pyramid
- Laplacian pyramid
*Two fundamental operations for pyramids*
- First operation blurs and samples the input
, - Second operation *interpolates* the blurred and sampled image to estimate the original
*Laplacian Error Pyramid*
- Error is difference between interpolated estimate and original
- Original signal can be recovered exactly by interpolating, then summing all the levels of the
error pyramid
- Can be represented very efficiently
*Useful for image coding/compression and progressive transmission*
*Gaussian pyramid useful for other tasks*
*Region Segmentation* - ANS-*Region segmentation*
*Background subtraction*
- Simple differences
- Statistical distance
- Multi-modal
*Morphology*
- Dilation to enlarge region
- Erosion to reduce region
- Closing (dilation + erosion)
- Opening (erosion + dilation)
*Region Growing*
- Connected components
- sequential
- recursive
*2D Shape* - ANS-*Given a 2-D binary shape, use representations/properties to characterize
the region*
- Recognition or matching
*Methods*
- Chain code
- Quadtree
- Medial axis
- Bounding box, extremal points
- Area, centroid
- Perimenter, compactness, circularity
- Signatures
- Moments
Hough Transform
- Used to identify/represent shapes in images by searching all possible parameters
*PCA and Face Recognition* - ANS-*Reduce high-dimensional input to lowerdimensional
"sub-space"*
- Beneficial for recognition
*PCA offers linear approximation to the sub-space
which can be reduced to only the major sub-space
