We are a fun project for which we need to compare two black and white bitmaps of two signatures and say that they Signature of individuals or not, because it is just two load bitmaps instead of data obtained from tablets, the approach for general signature recognition will be slightly different.
I'm thinking that the following steps will be required
- just crop the bitmap into the signature
- To make them rotate, Try to work on
- Change the shape to make the crop / rotated bitmap equal
- Analyze the signature inside (maybe breaking into a grid)
Do anyone have any ideas on this project? How to best perform rotation, comparison etc. Does anything look the same?
You might want to see: interesting pictures (for D) as well as image similarities Example of the method of comparison:
There are two main types of neural networks - supervision and unstable. SOMs are uncertain. Depending on your situation, you want to take a look at the supervised NN, and generally it is quite easy to implement for the most part.
No comments:
Post a Comment