This is the matlab implementation for BLPOC, Location based matching and DoG for reference points selection, the dataset used in this implementation is CASIA Palmprint database. For more detailed information about this techniques applied here, please refer to this paper: Performace Improvement of Phase-based Correspondence Matching for Palmprint Recognition.
Notice: Database prerocessing (ROI cropping, rotation, normalization) is not in this implementation.
You can start this code by typing:
[dis, blpoc] = matching('filename',mode)
filename
is actually a txt file, filled with lines of image paths in the following forms
path\to\image\xxx0.jpg path\to\image\xxx1.jpg
with a space in between representing two images that are gonna be matched together.
mode
specifies the approaches applied. 0
means location based method and 1
means applying DoG for reference points selection.
and this will return two vectors dis
the score gained from the locations and blpoc
for the average for BLPOC values of each corresponding image patches.
##Future work Since this is the only starting version of this implementation of the paper, further reconstruction and revison will be done.