# Feature Extraction and Descriptors

Below `[]`

in an argument list means an optional argument.

## Types

`ImageFeatures.Feature`

— Type`feature = Feature(keypoint, orientation = 0.0, scale = 0.0)`

The `Feature`

type has the keypoint, its orientation and its scale.

`ImageFeatures.Features`

— Type```
features = Features(boolean_img)
features = Features(keypoints)
```

Returns a `Vector{Feature}`

of features generated from the `true`

values in a boolean image or from a list of keypoints.

`ImageFeatures.Keypoint`

— Type```
keypoint = Keypoint(y, x)
keypoint = Keypoint(feature)
```

A `Keypoint`

may be created by passing the coordinates of the point or from a feature.

`ImageFeatures.Keypoints`

— Type```
keypoints = Keypoints(boolean_img)
keypoints = Keypoints(features)
```

Creates a `Vector{Keypoint}`

of the `true`

values in a boolean image or from a list of features.

`ImageFeatures.BRIEF`

— Type`brief_params = BRIEF([size = 128], [window = 9], [sigma = 2 ^ 0.5], [sampling_type = gaussian], [seed = 123])`

Argument | Type | Description |
---|---|---|

size | Int | Size of the descriptor |

window | Int | Size of sampling window |

sigma | Float64 | Value of sigma used for initial gaussian smoothing of image |

sampling_type | Function | Type of sampling used for building the descriptor (See BRIEF Sampling Patterns) |

seed | Int | Random seed used for generating the sampling pairs. For matching two descriptors, the seed used to build both should be same. |

`ImageFeatures.ORB`

— Type`orb_params = ORB([num_keypoints = 500], [n_fast = 12], [threshold = 0.25], [harris_factor = 0.04], [downsample = 1.3], [levels = 8], [sigma = 1.2])`

Argument | Type | Description |
---|---|---|

num_keypoints | Int | Number of keypoints to extract and size of the descriptor calculated |

n_fast | Int | Number of consecutive pixels used for finding corners with FAST. See [`fastcorners` ] |

threshold | Float64 | Threshold used to find corners in FAST. See [`fastcorners` ] |

harris_factor | Float64 | Harris factor `k` used to rank keypoints by harris responses and extract the best ones |

downsample | Float64 | Downsampling parameter used while building the gaussian pyramid. See [`gaussian_pyramid` ] in Images.jl |

levels | Int | Number of levels in the gaussian pyramid. See [`gaussian_pyramid` ] in Images.jl |

sigma | Float64 | Used for gaussian smoothing in each level of the gaussian pyramid. See [`gaussian_pyramid` ] in Images.jl |

`ImageFeatures.FREAK`

— Type`freak_params = FREAK([pattern_scale = 22.0])`

Argument | Type | Description |
---|---|---|

pattern_scale | Float64 | Scaling factor for the sampling window |

`ImageFeatures.BRISK`

— Type`brisk_params = BRISK([pattern_scale = 1.0])`

Argument | Type | Description |
---|---|---|

`pattern_scale` | `Float64` | Scaling factor for the sampling window |

## Corners and edges

`ImageFeatures.corner_orientations`

— Function```
orientations = corner_orientations(img)
orientations = corner_orientations(img, corners)
orientations = corner_orientations(img, corners, kernel)
```

Returns the orientations of corner patches in an image. The orientation of a corner patch is denoted by the orientation of the vector between intensity centroid and the corner. The intensity centroid can be calculated as `C = (m01/m00, m10/m00)`

where mpq is defined as -

``mpq = (x^p)(y^q)I(y, x) for each p, q in the corner patch``

The kernel used for the patch can be given through the `kernel`

argument. The default kernel used is a gaussian kernel of size `5x5`

.

`ImageCorners.fastcorners`

— Function`fastcorners(img, n, threshold) -> corners`

Performs FAST Corner Detection. `n`

is the number of contiguous pixels which need to be greater (lesser) than intensity + threshold (intensity - threshold) for a pixel to be marked as a corner. The default value for n is 12.

`Images.canny`

— Function`canny_edges = canny(img, (upper, lower), sigma=1.4)`

Performs Canny Edge Detection on the input image.

Parameters :

(upper, lower) : Bounds for hysteresis thresholding sigma : Specifies the standard deviation of the gaussian filter

**Example**

`imgedg = canny(img, (Percentile(80), Percentile(20)))`

`Images.phase`

— Function`phase(grad_x, grad_y) -> p`

Calculate the rotation angle of the gradient given by `grad_x`

and `grad_y`

. Equivalent to `atan(-grad_y, grad_x)`

, except that when both `grad_x`

and `grad_y`

are effectively zero, the corresponding angle is set to zero.

## BRIEF Sampling Patterns

`ImageFeatures.random_uniform`

— Function`sample_one, sample_two = random_uniform(size, window, seed)`

Builds sampling pairs using random uniform sampling.

`ImageFeatures.random_coarse`

— Function`sample_one, sample_two = random_coarse(size, window, seed)`

Builds sampling pairs using random sampling over a coarse grid.

`ImageFeatures.gaussian`

— Function`sample_one, sample_two = gaussian(size, window, seed)`

Builds sampling pairs using gaussian sampling.

`ImageFeatures.gaussian_local`

— Function`sample_one, sample_two = gaussian_local(size, window, seed)`

Pairs `(Xi, Yi)`

are randomly sampled using a Gaussian distribution where first `X`

is sampled with a standard deviation of `0.04*S^2`

and then the `Yi’s`

are sampled using a Gaussian distribution – Each `Yi`

is sampled with mean `Xi`

and standard deviation of `0.01 * S^2`

`ImageFeatures.center_sample`

— Function`sample_one, sample_two = center_sample(size, window, seed)`

Builds sampling pairs `(Xi, Yi)`

where `Xi`

is `(0, 0)`

and `Yi`

is sampled uniformly from the window.

## Feature Extraction

## Feature Description

`ImageFeatures.create_descriptor`

— Function```
desc, keypoints = create_descriptor(img, keypoints, params)
desc, keypoints = create_descriptor(img, params)
```

Create a descriptor for each entry in `keypoints`

from the image `img`

. `params`

specifies the parameters for any of several descriptors:

Some descriptors support discovery of the `keypoints`

from `fastcorners`

.

## Feature Matching

`ImageFeatures.hamming_distance`

— Function`distance = hamming_distance(desc_1, desc_2)`

Calculates the hamming distance between two descriptors.

`ImageFeatures.match_keypoints`

— Function`matches = match_keypoints(keypoints_1, keypoints_2, desc_1, desc_2, threshold = 0.1)`

Finds matched keypoints using the `hamming_distance`

function having distance value less than `threshold`

.

# Texture Matching

## Gray Level Co-occurence Matrix

`ImageFeatures.glcm`

— Function```
glcm = glcm(img, distance, angle, mat_size=16)
glcm = glcm(img, distances, angle, mat_size=16)
glcm = glcm(img, distance, angles, mat_size=16)
glcm = glcm(img, distances, angles, mat_size=16)
```

Calculates the GLCM (Gray Level Co-occurrence Matrix) of an image. The `distances`

and `angles`

arguments may be a single integer or a vector of integers if multiple GLCMs need to be calculated. The `mat_size`

argument is used to define the granularity of the GLCM.

`ImageFeatures.glcm_symmetric`

— Function```
glcm = glcm_symmetric(img, distance, angle, mat_size=16)
glcm = glcm_symmetric(img, distances, angle, mat_size=16)
glcm = glcm_symmetric(img, distance, angles, mat_size=16)
glcm = glcm_symmetric(img, distances, angles, mat_size=16)
```

Symmetric version of the `glcm`

function.

`ImageFeatures.glcm_norm`

— Function```
glcm = glcm_norm(img, distance, angle, mat_size)
glcm = glcm_norm(img, distances, angle, mat_size)
glcm = glcm_norm(img, distance, angles, mat_size)
glcm = glcm_norm(img, distances, angles, mat_size)
```

Normalised version of the `glcm`

function.

`ImageFeatures.glcm_prop`

— FunctionMultiple properties of the obtained GLCM can be calculated by using the `glcm_prop`

function which calculates the property for the entire matrix. If grid dimensions are provided, the matrix is divided into a grid and the property is calculated for each cell resulting in a height x width property matrix.

```
prop = glcm_prop(glcm, property)
prop = glcm_prop(glcm, height, width, property)
```

Various properties can be calculated like `mean`

, `variance`

, `correlation`

, `contrast`

, `IDM`

(Inverse Difference Moment), `ASM`

(Angular Second Moment), `entropy`

, `max_prob`

(Max Probability), `energy`

and `dissimilarity`

.

Missing docstring for `max_prob`

. Check Documenter's build log for details.

Missing docstring for `contrast`

. Check Documenter's build log for details.

Missing docstring for `ASM`

. Check Documenter's build log for details.

Missing docstring for `IDM`

. Check Documenter's build log for details.

Missing docstring for `glcm_entropy`

. Check Documenter's build log for details.

Missing docstring for `energy`

. Check Documenter's build log for details.

Missing docstring for `dissimilarity`

. Check Documenter's build log for details.

Missing docstring for `correlation`

. Check Documenter's build log for details.

Missing docstring for `glcm_mean_ref`

. Check Documenter's build log for details.

Missing docstring for `glcm_mean_neighbour`

. Check Documenter's build log for details.

Missing docstring for `glcm_var_ref`

. Check Documenter's build log for details.

Missing docstring for `glcm_var_neighbour`

. Check Documenter's build log for details.

## Local Binary Patterns

Missing docstring for `lbp`

. Check Documenter's build log for details.

Missing docstring for `modified_lbp`

. Check Documenter's build log for details.

Missing docstring for `direction_coded_lbp`

. Check Documenter's build log for details.

Missing docstring for `lbp_original`

. Check Documenter's build log for details.

Missing docstring for `lbp_uniform`

. Check Documenter's build log for details.

Missing docstring for `lbp_rotation_invariant`

. Check Documenter's build log for details.

Missing docstring for `multi_block_lbp`

. Check Documenter's build log for details.

# Misc

`ImageFeatures.HOG`

— Type`hog_params = HOG([orientations = 9], [cell_size = 8], [block_size = 2], [block_stride = 1], [norm_method = "L2-norm"])`

Histogram of Oriented Gradient (HOG) is a dense feature descriptor usually used for object detection. See "Histograms of Oriented Gradients for Human Detection" by Dalal and Triggs.

Parameters:

- orientations = number of orientation bins
- cell
*size = size of a cell is cell*size x cell_size (in pixels) - block
*size = size of a block is block*size x block_size (in terms of cells) - block_stride = stride of blocks. Controls how much adjacent blocks overlap.
- norm_method = block normalization method. Options: L2-norm, L2-hys, L1-norm, L2-sqrt.

`ImageFeatures.hough_transform_standard`

— Function```
lines = hough_transform_standard(
img_edges::AbstractMatrix;
stepsize=1,
angles=range(0,stop=pi,length=minimum(size(img))),
vote_threshold=minimum(size(img)) / stepsize -1,
max_linecount=typemax(Int))
```

Returns a vector of tuples corresponding to the tuples of (r,t) where r and t are parameters for normal form of line: `x * cos(t) + y * sin(t) = r`

`r`

= length of perpendicular from (1,1) to the line`t`

= angle between perpendicular from (1,1) to the line and x-axis

The lines are generated by applying hough transform on the image.

Parameters:

`img_edges`

= Image to be transformed (eltype should be`Bool`

)`stepsize`

= Discrete step size for perpendicular length of line`angles`

= List of angles for which the transform is computed`vote_threshold`

= Accumulator threshold for line detection`max_linecount`

= Maximum no of lines to return

**Example**

```
julia> using ImageFeatures
julia> img = fill(false,5,5); img[3,:] .= true; img
5×5 Array{Bool,2}:
false false false false false
false false false false false
true true true true true
false false false false false
false false false false false
julia> hough_transform_standard(img)
1-element Array{Tuple{Float64,Float64},1}:
(3.0, 1.5707963267948966)
```

`ImageFeatures.hough_circle_gradient`

— Function`circle_centers, circle_radius = hough_circle_gradient(img_edges, img_phase, radii; scale=1, min_dist=minimum(radii), vote_threshold)`

Returns two vectors, corresponding to circle centers and radius.

The circles are generated using a hough transform variant in which a non-zero point only votes for circle centers perpendicular to the local gradient. In case of concentric circles, only the largest circle is detected.

Parameters:

`img_edges`

= edges of the image`img_phase`

= phase of the gradient image`radii`

= circle radius range`scale`

= relative accumulator resolution factor`min_dist`

= minimum distance between detected circle centers`vote_threshold`

= accumulator threshold for circle detection

`canny`

and `phase`

can be used for obtaining img*edges and img*phase respectively.

**Example**

```
julia> using Images, ImageFeatures, FileIO, ImageView
julia> img = load(download("http://docs.opencv.org/3.1.0/water_coins.jpg"));
julia> img = Gray.(img);
julia> img_edges = canny(img, (Percentile(99), Percentile(80)));
julia> dx, dy=imgradients(img, KernelFactors.ando5);
julia> img_phase = phase(dx, dy);
julia> centers, radii = hough_circle_gradient(img_edges, img_phase, 20:30);
julia> img_demo = Float64.(img_edges); for c in centers img_demo[c] = 2; end
julia> imshow(img_demo)
```