26-06-2012, 04:33 PM
SIGNATURE RECOGNITION USING IMAGE PROCESSING
[attachment=25636]
INTRODUCTION:
Signature is the characteristic of the particular person & hence used globally for identifying a person, validity of the documents signed, banking etc. Up till now, in banks where signature of a person is the basic code for transaction, the validity of the signature is generally checked by a man. Signature recognition applications simulate the ability of a man to recognize a signature from the standard signature he has. A system which recognizes the signature is implemented.
WHAT IS SIGNATURE RECOGNITION?
Biometric Identification by Automatically scanning a person's signature and matching it electronically against a library of known signatures. The image of the written text may be sensed "off line" from a piece of paper by optical scanning. Alternatively, the movements of the pen tip may be sensed "on line", for example by a pen-based computer screen surface.
INPUT SECTION:
There are two inputs to this section. One input is used for creating new customer account in bank, while other being used for importing the signature for verification using media such as scanner, web-cam. For importing the signature there must be a copy of standardized signature for the corresponding user account number. If the customer opens new account then the related information is stored in database with some modifications according to the requirement. NO one can interfere with the database of the software as this facility is available only to the supervisor of program. This is done by scanning the required signatures or via an electronic writing pad
PRE-PROCESSING BLOCK:
The hardware being used for implementing the Software offers certain limitations as we are going to use the scanner, monitor and VGA of the computer which have finite resolution (not very high) and capabilities, which ultimately results in the loss of data at the input side of the program. Hence, pre-processing becomes necessary.
Colour normalisation: In recognition process colour normalization is very essential. The signatures can be done on any paper with different colour inks, hence colour normalization is used in which we make the signature BLACK and the background is made WHITE. Due to the image scaling process the picture becomes blocky; hence we perform a smoothening operation, which may introduce grey shades to the signature. Hence the colour normalization is re initiated. A histogram is a convenient way to describe the data. To form a histogram; the data from a single class are grouped into intervals. Over each interval, a vertical rectangle is drawn, with its area proportional to the number of data points falling into that interval. It basically gives us the information about the number of pixels in an image of particular intensity/luminosity.
Finding the centroid of an image:
During the process of EX-Oring i.e. checking the signature, it is required that the program must try to overlap the test signature in such a fashion that maximum no. of pixels are marked as correct. Though man can do this difficult thing very easily; to find out the resonating position for the test signature is quite difficult, as trial and error method though can give maximum resonance is very much time consuming. Another method that can be used is by finding the extreme of the signature & overlaps the test signature (duplicate signature) over the standard signature using the extremes.
But as many of the signatures have changing extreme parts this method may result drastically false results in some cases. Hence, we have implemented the centroid method in which the test signature is moved in such fashion that centroids of the both signatures are same.
[attachment=25636]
INTRODUCTION:
Signature is the characteristic of the particular person & hence used globally for identifying a person, validity of the documents signed, banking etc. Up till now, in banks where signature of a person is the basic code for transaction, the validity of the signature is generally checked by a man. Signature recognition applications simulate the ability of a man to recognize a signature from the standard signature he has. A system which recognizes the signature is implemented.
WHAT IS SIGNATURE RECOGNITION?
Biometric Identification by Automatically scanning a person's signature and matching it electronically against a library of known signatures. The image of the written text may be sensed "off line" from a piece of paper by optical scanning. Alternatively, the movements of the pen tip may be sensed "on line", for example by a pen-based computer screen surface.
INPUT SECTION:
There are two inputs to this section. One input is used for creating new customer account in bank, while other being used for importing the signature for verification using media such as scanner, web-cam. For importing the signature there must be a copy of standardized signature for the corresponding user account number. If the customer opens new account then the related information is stored in database with some modifications according to the requirement. NO one can interfere with the database of the software as this facility is available only to the supervisor of program. This is done by scanning the required signatures or via an electronic writing pad
PRE-PROCESSING BLOCK:
The hardware being used for implementing the Software offers certain limitations as we are going to use the scanner, monitor and VGA of the computer which have finite resolution (not very high) and capabilities, which ultimately results in the loss of data at the input side of the program. Hence, pre-processing becomes necessary.
Colour normalisation: In recognition process colour normalization is very essential. The signatures can be done on any paper with different colour inks, hence colour normalization is used in which we make the signature BLACK and the background is made WHITE. Due to the image scaling process the picture becomes blocky; hence we perform a smoothening operation, which may introduce grey shades to the signature. Hence the colour normalization is re initiated. A histogram is a convenient way to describe the data. To form a histogram; the data from a single class are grouped into intervals. Over each interval, a vertical rectangle is drawn, with its area proportional to the number of data points falling into that interval. It basically gives us the information about the number of pixels in an image of particular intensity/luminosity.
Finding the centroid of an image:
During the process of EX-Oring i.e. checking the signature, it is required that the program must try to overlap the test signature in such a fashion that maximum no. of pixels are marked as correct. Though man can do this difficult thing very easily; to find out the resonating position for the test signature is quite difficult, as trial and error method though can give maximum resonance is very much time consuming. Another method that can be used is by finding the extreme of the signature & overlaps the test signature (duplicate signature) over the standard signature using the extremes.
But as many of the signatures have changing extreme parts this method may result drastically false results in some cases. Hence, we have implemented the centroid method in which the test signature is moved in such fashion that centroids of the both signatures are same.