detectKAZEFeatures

Detect KAZE features

Syntax

points = detectKAZEFeatures(I)

points = detectKAZEFeatures(I,Name=Value)

Description

points = detectKAZEFeatures(I) returns a KAZEPoints object containing information about KAZE keypoints detected in a 2-D grayscale or binary image. The function uses nonlinear diffusion to construct a scale space for the given image. It then detects multiscale corner features from the scale space.

points = detectKAZEFeatures(I,Name=Value) specifies options using one or more name-value arguments in addition to the previous syntax. For example, detectKAZEFeatures(I,Threshold=0.003) sets the threshold to exclude less significant local extrema to 0.003.

Examples

Detect KAZE Point Features in Image

Read an image.

I = imread("cameraman.tif");

Detect KAZE points in the image.

points = detectKAZEFeatures(I);

Plot the 20 strongest points.

imshow(I)
hold on
plot(selectStrongest(points,20))
hold off

Detect KAZE Features and Display Specific Points

Read an image into the workspace.

I = imread("cameraman.tif");

Detect KAZE features in the image.

points = detectKAZEFeatures(I);

Select and display the last 5 points detected.

imshow(I);
hold on;
plot(points(end-4:end));
hold off;

Input Arguments

I — Input image

M-by-N 2-D grayscale or binary image

Input image, specified as an M-by-N 2-D grayscale or binary image.

Data Types: single | double | int16 | uint8 | uint16 | logical

Name-Value Arguments

Specify optional pairs of arguments as Name1=Value1,...,NameN=ValueN, where Name is the argument name and Value is the corresponding value. Name-value arguments must appear after other arguments, but the order of the pairs does not matter.

Example: detectKAZEFeatures(I,Threshold=0.003) sets the threshold to exclude less significant local extrema to 0.003.

Diffusion — Method to compute conductivity

"region" (default) | "sharpedge" | "edge"

Method to compute conductivity, specified as "region", "sharpedge", or "edge". The computation is based on first-order derivatives of a layer in scale space.

Method	Selected Features
"region"	Large regions. Uses the Perona and Malik conductivity coefficient, 1/(1 + dL^2/k^2).
"sharpedge"	High-contrast edges. Uses the Perona and Malik conductivity coefficient, exp(-|dL|^2/k^2).
"edge"	Smoothing on both sides of an edge rather than across it. Uses the Weickert conductivity coefficient.

Threshold — Local extrema

0.0001 (default) | scalar

Local extrema, specified as a scalar greater than or equal to 0. Increase this value to exclude less significant local extrema.

NumOctaves — Multiscale detection factor

3 (default) | positive integer

Multiscale detection factor, specified as a positive integer. Increase this value to detect larger features. To disable multiscale detection, set NumOctaves to 1. When you set the value to 1, the function performs the detection at the scale of the input image. Recommended values are between 1 and 4.

NumScaleLevels — Scale levels

4 (default) | integer

Scale levels, specified as an integer in the range [3,10]. Increase this value to achieve smoother scale changes. Increasing this value also provides additional intermediate scales between octaves.

ROI — Rectangular region size

[1 1 size(I,2) size(I,1)] (default) | 4-element vector

Rectangular region size for corner detection, specified as a 4-element vector in the format [y x width height]. The [y x ] values are measured from the upper left corner of the rectangle.

Output Arguments

points — KAZE points object

KAZEpoints object

KAZE points, returned as a KAZEPoints object. The object contains information about the point features detected in the 2-D grayscale or binary input image.

References

[1] Alcantarilla, P.F., A. Bartoli, and A.J. Davison. "KAZE Features." ECCV 2012, Part VI, LNCS 7577. 2012, p. 214

Extended Capabilities

C/C++ Code Generation

Generate C and C++ code using MATLAB® Coder™.

Usage notes and limitations:

• Generates portable C code using a C++ compiler that links to OpenCV libraries. See Portable C Code Generation for Functions That Use OpenCV Library.

Version History

Introduced in R2017b