shs - Sample Hough Space. A tool to sample the Motion Hough Space. The sampled data can be used to visualize the shape of the Hough Space, using any package to display 3D data (for example xprism). The program also returns the position of the global minimum and estimate confidence matrix.
NAME
shs
shs
-s image_sequence_name
-x x_coordinate_of_ROI
-y y_coordinate_of_ROI
-h horizontal_size_of_ROI
-v vertical_size_of_ROI
[
-r resolution_in_the_HS
]
[
-f passes_of_binomial_filter
]
[
-m motion_threshold
]
[
-k kernel_type
]
[
-a vx_min_value
]
[
-A vx_max_value
]
[
-b vy_min_value
]
[
-B vy_max_value
]
[
-o output_filename
]
[
-i interactive_flag
]
[
-U print_usage
]
The shs program samples the Hough Space (HS) for a given sequence (two frames saved in .seq format)
and saves results in text files. The HS is sampled with the pre-specified resolution, and for
motion parameters values from the specified intervals. The two parameter motion model is used.
A set of kernels is available, including:
Absolute, Quadratic, Hampel_1_2_1, Huber min-max, Tukey. The
optimal kernel
depends on the type of sequence and noise level and distribution.
- -s image_sequence_name
- Specifies the file name of the sequence (stored in the sequence [.seq] format).
- -f number_of_passes_of_binomial_filter (default 3)
- Defines the number of passes of the binomial filter used to low-pass filter the sequence.
- -m motion_threshold (default 0)
- Defines the threshold on the Frame Difference. Pixels for which the Frame Difference
is less then the threshold do not contribute to the Hough Space
- -x x_coord
- x coordinate of the top right corner of the Region of Interest (ROI)
- -y y_coord
- y coordinate of the top right corner of the ROI
- -h horizontal_size
- horizontal size of the Region of Interest
- -v vertical_size
- vertical size of the Region of Interest
- -r resolution_in_the_HS (default 1 pel)
- Specifies the resolution in the Hough (Motion Parameter) Space
- -k kernel_type (default 0)
- Specifies the kernel type:
0 - Absolute
1 - Quadratic
2 - Huber min-max
3 - Hampel 1-2-1
4 - Tukey
5 - Normalized Correlation
- -a vx_min -A vx_max (default [-8,8])
-
- -b vy_min -B vy_max (default [-8,8])
-
- -o output_filename (default hough)
- Write samples of the Hough Space into a text files. Three file are created:
hough_space.dat
- total support for motion parameters from the ROI
hough_mean.dat
- mean value of support form the region ROI
hough_std.dat
- standard deviation of support within the ROI
- -U
- Prints usage
Example 1 (running for the command line)
shs -s is4.seq -x20 -y20 -h32 -v32 -f 3 -r 0.1 -a -7.0 -A -3.0 -b -2.0 -B 2.0 -o is4_hough
Sample Hough Space for sequence <is4.seq>. Low-pass filter reference and consecutive frames
using Binomial filter (3 passes). Sample Hough Space for Region of Interest with upper right
corner at pixel (20,20) and horizontal and vertical size of 32 pixels. Region of interest
in the Hough Space is defined as [-7.0, -3.0] for Vx and [-2.0, 2.0] for Vy. Resolution in
the Parameter space is 0.1 pel. Save results to files with steam is4_hough (is4_hough_space.dat,
is4_hough_mean.dat, is4_hough_std.dat are created).
Example 2 (interactive use)
shs -i
Interactive run. Program prompts the user for the values of run parameters.
RHflow2D, RHflow4D, RHglob2D, RHglob4Dr.
For the definition of the Hough Space see:
[1] M. Bober and J. Kittler, Robust motion analysis, CVPR1994, pp 947--952, IEEE Computer Society Press.
[2] M. Bober and J. Kittler, Estimation of general multimodal motion: an approach based on
robust statistics and Hough transform, Image and Vision Computing, Volume 12, Number 12, 1994,
pp 661--668
During sampling procedure, the block displacement is not checked against valid block
positions. If displacement range move block out of the image
a segmentation error may result.
This program is copyrighted, see the Copy-Right Notice.
Miroslaw Bober, Vision Speech and Signal Processing Group,
University of Surrey, Guildford, Surrey, GU2 5XH, United Kingdom.
M.Bober@ee.surrey.ac.uk
03-Feb-98. Automatically converted by man2html,
written by G.Matas (g.matas@ee.surrey.ac.uk)