GaborFeatureExtract: Gabor Feature Extraction

Description Usage Details Methods Methods References Examples

Description

Gabor Feature Extraction

Gabor Feature Extraction

Usage

1
# init <- GaborFeatureExtract$new()

Details

In case of an RGB image (3-dimensional) one can use the rgb_2gray() to convert the image to a 2-dimensional one

I added the option downsample_gabor to the original matlab code based on the following question on stackoverflow : https://stackoverflow.com/questions/49119991/feature-extraction-with-gabor-filters

Methods

GaborFeatureExtract$new()
--------------
gabor_filter_bank(scales, orientations, gabor_rows, gabor_columns, plot_data = FALSE)
--------------
gabor_feature_extraction(image, scales, orientations, gabor_rows, gabor_columns, downsample_gabor = FALSE, plot_data = FALSE, downsample_rows = NULL, downsample_cols = NULL, normalize_features = FALSE, threads = 1, vectorize_magnitude = TRUE)
--------------
gabor_feature_engine(img_data, img_nrow, img_ncol, scales, orientations, gabor_rows, gabor_columns, downsample_gabor = FALSE, downsample_rows = NULL, downsample_cols = NULL, normalize_features = FALSE, threads = 1, verbose = FALSE)
--------------
plot_gabor(real_matrices, margin_btw_plots = 0.15, thresholding = FALSE)
--------------
plot_multi_images(list_images, par_ROWS, par_COLS)
--------------

Methods

Public methods


Method new()

Usage
GaborFeatureExtract$new()

Method gabor_filter_bank()

Usage
GaborFeatureExtract$gabor_filter_bank(
  scales,
  orientations,
  gabor_rows,
  gabor_columns,
  plot_data = FALSE
)
Arguments
scales

a numeric value. Number of scales (usually set to 5) ( gabor_filter_bank function )

orientations

a numeric value. Number of orientations (usually set to 8) ( gabor_filter_bank function )

gabor_rows

a numeric value. Number of rows of the 2-D Gabor filter (an odd integer number, usually set to 39 depending on the image size) ( gabor_filter_bank function )

gabor_columns

a numeric value. Number of columns of the 2-D Gabor filter (an odd integer number, usually set to 39 depending on the image size) ( gabor_filter_bank function )

plot_data

either TRUE or FALSE. If TRUE then data needed for plotting will be returned ( gabor_filter_bank, gabor_feature_extraction functions )


Method gabor_feature_extraction()

Usage
GaborFeatureExtract$gabor_feature_extraction(
  image,
  scales,
  orientations,
  gabor_rows,
  gabor_columns,
  downsample_gabor = FALSE,
  plot_data = FALSE,
  downsample_rows = NULL,
  downsample_cols = NULL,
  normalize_features = FALSE,
  threads = 1,
  verbose = FALSE,
  vectorize_magnitude = TRUE
)
Arguments
image

a 2-dimensional image of type matrix ( gabor_feature_extraction function )

scales

a numeric value. Number of scales (usually set to 5) ( gabor_filter_bank function )

orientations

a numeric value. Number of orientations (usually set to 8) ( gabor_filter_bank function )

gabor_rows

a numeric value. Number of rows of the 2-D Gabor filter (an odd integer number, usually set to 39 depending on the image size) ( gabor_filter_bank function )

gabor_columns

a numeric value. Number of columns of the 2-D Gabor filter (an odd integer number, usually set to 39 depending on the image size) ( gabor_filter_bank function )

downsample_gabor

either TRUE or FALSE. If TRUE then downsampling of data will take place. The downsample_rows and downsample_cols should be adjusted accordingly. Downsampling does not affect the output plots but the output gabor_features ( gabor_feature_extraction function )

plot_data

either TRUE or FALSE. If TRUE then data needed for plotting will be returned ( gabor_filter_bank, gabor_feature_extraction functions )

downsample_rows

either NULL or a numeric value specifying the factor of downsampling along rows ( gabor_feature_extraction function )

downsample_cols

either NULL or a numeric value specifying the factor of downsampling along columns ( gabor_feature_extraction function )

normalize_features

either TRUE or FALSE. If TRUE then the output gabor-features will be normalized to zero mean and unit variance ( gabor_feature_extraction function )

threads

a numeric value specifying the number of threads to use ( gabor_feature_extraction function )

verbose

either TRUE or FALSE. If TRUE then information will be printed in the console ( gabor_feature_extraction, gabor_feature_engine functions )

vectorize_magnitude

either TRUE or FALSE. If TRUE the computed magnitude feature will be returned in the form of a vector, otherwise it will be returned as a list of matrices ( gabor_feature_extraction function )


Method gabor_feature_engine()

Usage
GaborFeatureExtract$gabor_feature_engine(
  img_data,
  img_nrow,
  img_ncol,
  scales,
  orientations,
  gabor_rows,
  gabor_columns,
  downsample_gabor = FALSE,
  downsample_rows = NULL,
  downsample_cols = NULL,
  normalize_features = FALSE,
  threads = 1,
  verbose = FALSE
)
Arguments
img_data

a numeric matrix specifying the input data (gabor_feature_engine function)

img_nrow

an integer specifying the number of rows of the input matrix (gabor_feature_engine function)

img_ncol

an integer specifying the number of columns of the input matrix (gabor_feature_engine function)

scales

a numeric value. Number of scales (usually set to 5) ( gabor_filter_bank function )

orientations

a numeric value. Number of orientations (usually set to 8) ( gabor_filter_bank function )

gabor_rows

a numeric value. Number of rows of the 2-D Gabor filter (an odd integer number, usually set to 39 depending on the image size) ( gabor_filter_bank function )

gabor_columns

a numeric value. Number of columns of the 2-D Gabor filter (an odd integer number, usually set to 39 depending on the image size) ( gabor_filter_bank function )

downsample_gabor

either TRUE or FALSE. If TRUE then downsampling of data will take place. The downsample_rows and downsample_cols should be adjusted accordingly. Downsampling does not affect the output plots but the output gabor_features ( gabor_feature_extraction function )

downsample_rows

either NULL or a numeric value specifying the factor of downsampling along rows ( gabor_feature_extraction function )

downsample_cols

either NULL or a numeric value specifying the factor of downsampling along columns ( gabor_feature_extraction function )

normalize_features

either TRUE or FALSE. If TRUE then the output gabor-features will be normalized to zero mean and unit variance ( gabor_feature_extraction function )

threads

a numeric value specifying the number of threads to use ( gabor_feature_extraction function )

verbose

either TRUE or FALSE. If TRUE then information will be printed in the console ( gabor_feature_extraction, gabor_feature_engine functions )


Method plot_gabor()

Usage
GaborFeatureExtract$plot_gabor(
  real_matrices,
  margin_btw_plots = 0.65,
  thresholding = FALSE
)
Arguments
real_matrices

a list of 3-dimensional arrays. These arrays correspond to the real part of the complex output matrices ( plot_gabor function )

margin_btw_plots

a float between 0.0 and 1.0 specifying the margin between the multiple output plots ( plot_gabor function )

thresholding

either TRUE or FALSE. If TRUE then a threshold of 0.5 will be used to push values above 0.5 to 1.0 ( similar to otsu-thresholding ) ( plot_gabor function )


Method plot_multi_images()

Usage
GaborFeatureExtract$plot_multi_images(list_images, par_ROWS, par_COLS)
Arguments
list_images

a list containing the images to plot ( plot_multi_images function )

par_ROWS

a numeric value specifying the number of rows of the plot-grid ( plot_multi_images function )

par_COLS

a numeric value specifying the number of columns of the plot-grid ( plot_multi_images function )


Method clone()

The objects of this class are cloneable with this method.

Usage
GaborFeatureExtract$clone(deep = FALSE)
Arguments
deep

Whether to make a deep clone.

References

https://github.com/mhaghighat/gabor

https://stackoverflow.com/questions/20608458/gabor-feature-extraction

https://stackoverflow.com/questions/49119991/feature-extraction-with-gabor-filters

Examples

 1
 2
 3
 4
 5
 6
 7
 8
 9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
66
67
68
69
70
71
72
73
74
75
76
77
library(OpenImageR)

init_gb = GaborFeatureExtract$new()

# gabor-filter-bank
#------------------

gb_f = init_gb$gabor_filter_bank(scales = 5, orientations = 8, gabor_rows = 39,

                                 gabor_columns = 39, plot_data = TRUE)


# plot gabor-filter-bank
#-----------------------

plt_f = init_gb$plot_gabor(real_matrices = gb_f$gabor_real, margin_btw_plots = 0.65,

                           thresholding = FALSE)


# read image
#-----------

pth_im = system.file("tmp_images", "car.png", package = "OpenImageR")

im = readImage(pth_im) * 255


# gabor-feature-extract
#----------------------

# gb_im = init_gb$gabor_feature_extraction(image = im, scales = 5, orientations = 8,

#                                          downsample_gabor = TRUE, downsample_rows = 3,

#                                          downsample_cols = 3, gabor_rows = 39, gabor_columns = 39,

#                                          plot_data = TRUE, normalize_features = FALSE,

#                                          threads = 6)


# plot real data of gabor-feature-extract
#----------------------------------------

# plt_im = init_gb$plot_gabor(real_matrices = gb_im$gabor_features_real, margin_btw_plots = 0.65,

#                             thresholding = FALSE)


# feature generation for a matrix of images (such as the mnist data set)
#-----------------------------------------------------------------------

ROWS = 13; COLS = 13; SCAL = 3; ORIEN = 5; nrow_mt = 500; im_width = 12; im_height = 15

set.seed(1)
im_mt = matrix(sample(1:255, nrow_mt * im_width * im_height, replace = TRUE), nrow = nrow_mt,

                      ncol = im_width * im_height)

# gb_ex = init_gb$gabor_feature_engine(img_data = im_mt, img_nrow = im_width, img_ncol = im_height,

#                                      scales = SCAL, orientations = ORIEN, gabor_rows = ROWS,

#                                      gabor_columns = COLS, downsample_gabor = FALSE,

#                                      downsample_rows = NULL, downsample_cols = NULL,

#                                      normalize_features = TRUE, threads = 1, verbose = FALSE)


# plot of multiple image in same figure
#---------------------------------------

list_images = list(im, im, im)

plt_multi = init_gb$plot_multi_images(list_images, par_ROWS = 2, par_COLS = 2)

OpenImageR documentation built on Sept. 10, 2021, 5:08 p.m.