/numberplatesCH

Detections of Swiss car number plates

Primary LanguageMATLABMIT LicenseMIT

numberplatesCH

Detections of Swiss car number plates

Original Script by: https://mathworks.com/matlabcentral/fileexchange/54456-licence-plate-recognition

This MATLAB Solution provides image detection of Swiss car number plates. The trainingfile can be trained to detect other images aswell.

Training

If you want to train the algorithm to your needs you can do so. The reference file is 'imgfildata.mat'. All image are stored in a two dimensional cell array. (black and white, 1 or 0). To "train" the 'imgfildata.mat' you shall use the 'training_imgfildata.m' script.

Images have to be two dimensional to store them into 'imgfildata.mat'. You can use our script to detect characters and automatticaly invert them and store then to a path with the file extension you need. Look at 'learning_edited.m' for that.

Training Folder

All data related to training is there. 'letters_numbers' folder is used for a reference when executing 'training_imgfildata.m'
Make sure your images are inverted and only contain black and white (0 and 1). Also the size must be 24x42px (can be adjusted).
The filename is used to write the desired letter into the 'imgfildata.mat' file. If you name the file 'L' that means that the file named 'L' is your reference for a 'L'.

Files

number_plate_det.m -> is the actual detection script
Training\imgfildata.mat -> reference file with trained characters
Training\training_imgfildata.m -> script to write into imgfildata.mat
learning_edited.m -> script to parse all characters from an image, invert them and store them individually

"Kennzeichen" Folder

You'll find some sample files there.

Compile with MATLAB MCC

If you want to use the project as an executable you can compile it yourself with mcc. Run the following in matlab
mcc -m -I 'C:\matlab source\compiled' -d 'C:\matlab source\exe' number_plate_det.m