An implementation of Distinctive image features from scale-invariant keypoints, created by David Lowe. This is a term project for "Advanced Topics in Medical Image Analysis" course at Middle East Technical University.
Read an image in MatLab and convert it into gray scale image then use it as input for SIFT function.
SIFT(Image, Octaves, Scales, Sigma):Main function takes gray scale image, number of octaves, number of scales per octaves and initial value for sigma. Returns key points inside a cell array.
Example:
image = imread('image.jpg');
image = double(rgb2gray(image));
keyPoints = SIFT(image,3,5,1.6);
Key points created in the process are objects. All of the key points are returned in a cell array. Each key point contains:
coordinates():returns [x, y] coordinate of the key point on image.direction():Returns general Direction of the key point.magnitude():Returns the magnitude of general direction vector.octave():Returns number of the octave which the key point extracted from.scale():Returns sigma value which the image is convolved with.descriptor():Returns a vector containing the descriptor.
This function's main purpose is to illustrate the keypoints on the image. The function SIFTKeypointVisualizer will be called after extracting keypoints.
Example:
image = SIFTKeypointVisualizer(image,keyPoints);
imshow(uint8(image));
This code is dependent on some functions within Image Processing Toolbox version 9.2.
This code should be compatible with any MatLab newer than R2015a.
All the rights to "Distinctive image features from scale-invariant keypoints" are reserved for University of British Columbia. A license must be obtained from the University of British Columbia for any commercial applications.
The patent is expired on March 6, 2020, and the algorithm is now - possibly - public knowledge.
For further information please contact zabardast.amin@metu.edu.tr or kilic.ozkan@metu.edu.tr.