18cse390t U1 s7 Slo1 Content
Uploaded by
Deepa S18cse390t U1 s7 Slo1 Content
Uploaded by
Deepa S18CSE390T
Computer Vision
U1 - S7 - SLO1 - Linear Filtering
Linear Filtering
• Locally adaptive histogram equalization is an example of a
neighborhood operator or local operator, which uses a collection of
pixel values in the vicinity of a given pixel to determine its final output
value
Neighborhood filtering (convolution): The image on the left is convolved with the filter in the middle to
yield the image on the right. The light blue pixels indicate the source neighborhood for the light green
destination pixel
Linear Filtering
• In addition to performing local tone adjustment, neighborhood
operators can be used to filter images to
• add soft blur,
• sharpen details,
• accentuate edges, or
• remove noise
• We will discuss linear filtering operators, which involve fixed weighted
combinations of pixels in small neighborhoods
Linear Filtering
• Some neighborhood operations: (a) original image; (b) blurred; (c)
sharpened; (d) smoothed with edge-preserving filter; (e) binary image; (f)
dilated; (g) distance transform; (h) connected components.
Linear Filtering
• The most widely used type of neighborhood operator is a linear filter,
where an output pixel’s value is a weighted sum of pixel values within
a small neighborhood N
• The entries in the weight kernel or mask h(k, l) are often called the
filter coefficients.
• The above correlation operator can be more compactly notated as
g = fh
Linear Filtering
• A common variant on this formula is
• where the sign of the offsets in f has been reversed, This is called the
convolution operator, g = f*h
• h is then called the impulse response function
Linear Filtering
• Padding (border effects)
• The correlation produces a result that is smaller than the original
image, which may not be desirable in many applications.
• This is because the neighborhoods of typical correlation and
convolution operations extend beyond the image boundaries near the
edges, and so the filtered images suffer from boundary effects
• To deal with this, a number of different padding or extension modes
have been developed for neighborhood operations
Linear Filtering
• Padding (border effects)
• zero: set all pixels outside the source image to 0 (a good choice for alpha-
matted cutout images)
• constant (border color): set all pixels outside the source image to a
specified border value
• clamp (replicate or clamp to edge): repeat edge pixels indefinitely
• (cyclic) wrap (repeat or tile): loop “around” the image in a “toroidal”
configuration
• mirror: reflect pixels across the image edge
• extend: extend the signal by subtracting the mirrored version of the signal
from the edge pixel value
Separable filtering
• The process of performing a convolution requires K2 (multiply-add)
operations per pixel, where K is the size (width or height) of the
convolution kernel
Separable filtering
• It is easy to show that the two-dimensional kernel K corresponding to
successive convolution with a horizontal kernel h and a vertical kernel
v is the outer product of the two kernels,
• How can we tell if a given kernel K is indeed separable?
• This can often be done by inspection or by looking at the analytic
form of the kernel.
• A more direct method is to treat the 2D kernel as a 2D matrix K and to
take its singular value decomposition (SVD),
Band-pass and steerable filters
• The Sobel and corner operators are simple examples of band-pass
and oriented filters.
• More sophisticated kernels can be created by first smoothing the
image with a (unit area) Gaussian filter,
and then taking the first or second derivatives
• Such filters are known collectively as band-pass filters, since they
filter out both low and high frequencies
Band-pass and steerable filters
• The (undirected) second derivative of a two-dimensional image is
known as the Laplacian operator
• Blurring an image with a Gaussian and then taking its Laplacian is
equivalent to convolving directly with the Laplacian of Gaussian (LoG)
filter which has certain nice scale-space properties
Band-pass and steerable filters
Second-order steerable filter (Freeman1992) © 1992 IEEE:
(a) original image of Einstein;
(b) orientation map computed from the second-order oriented energy;
(c) original image with oriented structures enhanced
Band-pass and steerable filters
Fourth-order steerable filter (Freeman and Adelson1991) © 1991 IEEE:
(a)test image containing bars (lines) and step edges at different
orientations;
(b)average oriented energy;
(c)dominant orientation;
(d)oriented energy as a function of angle (polar plot)