11-06-2013, 03:35 PM
SIGNATURE AUTHENTICATION
SIGNATURE AUTHENTICATION.pdf (Size: 203.55 KB / Downloads: 95)
ABSTRACT
Signature can be seen as an individual characteristic of a person which, if modeled with precision can be used for his/her validation. An automated signature authentication technique saves valuable time and money. The paper is primarily focused on skilled forgery detection. It emphasizes on the extraction of the critical regions which are more prone to mistakes and matches them following a modular graph matching approach. The technique is robust and takes care of the inevitable intra-personal variations. The results show significant improvement over other approaches for detecting skilled forgery.
INTRODUCTION
A handwritten signature as a behavioural biometric is the mean accepted method to declare someone's identity. Many documents necessitate a handwritten signature. In general, there are two ways to process the signature sample. The first is on-line, where the image is captured directly as handwriting trajectory. The second is off-line, in which we use a digitizer in order to acquire a digital image.
Signature Verification[1][2][3][7] is the process of recognizing an individual’s handwritten signatures. Signatures have been by far the most popular means for establishing the authenticity of individuals. Signature authentication[1][2][3][7] offers a quick, simple and cost effective means for validating the authenticity of a document by determining the difference between an original signature and a counterfeit one.
This project takes on the task of comparing signature which are store as image files in a folder. The purpose is to provide easy and fast matching procedure with highly accurate search results. This paper covers on offline verification of signature. It authenticates signature based on certain traits such as critical region extraction and matching.
The front end used is Matlab® (Mathworks IncLtd). MATLAB solutions are expressed in familiar mathematical notation. Typical uses includes: Math and computation, Algorithm development, Data acquisition, Modelling, Simulation, and prototyping, Data analysis, exploration, and visualization, Scientific and engineering graphics, Application development, including graphical user interface building,MATLAB is an interactive system whose basic data element is an array that does not require dimensioning. This allows formulating solutions to many technical computing problems, especially those involving matrix representations, in a fraction of the time it would take to write a program in a scalar non interactive language such as C or FORTRAN.
ALGORITHM DESIGN
The input image goes through the following procedure before judging that the two signatures are accurate or not. The steps are Binarization ,Noise Removal, Rotation, Thinning, Critical point
extraction, Critical region matching and Verification. Each step is mentioned in a separate section starting from section 2.1 to to 2.7. Figure 1 shows the approach in form of a diagrammatic representation.
BINARIZATION
The main components of binarization include statistical analysis of images, determination of a threshold value based upon the statistical results and applying the threshold value to gray-lcvel images. Statistical analysis of gray- level images may include determination of mean, variance. Standard deviation, contrast stretch, histogram etc. or it can be a combination of any of these Determination of a threshold value is very much important and perhaps the most sensitive part of any image binarization scheme because a wrong value of threshold may result in losing some image information (an object can be considered as part of background and vice versa). Moreover, the value of threshold should be sensitive with the overall contrast stretch of the corresponding image. The threshold value is applied to the image to represent it in I-bit after the proper determination of a threshold limit.
The simplest method to convert a gray scale image into a binary image is to compute mean of the image and set the value of mean as the threshold for binarization. But this approach has many shortcomings, as it may not take care about the features and objects in the image properly. This is due to the fact that the mean of an image may be disturbed drastically by the addition of noise pixels or very few numbers of pixels having the intensity close to any of the boundaries of gray scale.
NOISE REMOVAL
Once we have binarized an image, the noise components must be removed. Small components (pixel_size<5-10pixels) are removed by using a simple morphological filter. Assumption that the signature content would be more prevalent in the image, the image is passed though a low pass filter to eliminate the low frequency noise components. The filtering is done by using 2-D convolution with a 5X5 Unity matrix. The results are illustrated in figure 2a and 2b. This process converts the binary image into a gray scale image. The image thus obtained is further binarized sing a strict estimated threshold. We use Niblack’s Algorithm [4] for this.
ROTATION
The accuracy of the results is largely dependent on the rotation algorithm used for orientation correction. The rotation algorithm should be robust and must produce the same results for images taken from the same user. The rotation algorithm rotate-image is given below. The binarized and noise cleaned signature image is input to the algorithm. We use the bottom pixels of a signature image as a template to fit an orientation line through them using the polyfit function of Matlab® (Mathworks Inc Ltd). The polyfit Function is further explained in Section 3.1. Finally, the cp2tranform () function produces a projective transformation of the input image, using the slope of the orientation line as a guiding parameter. Experimentally, we found that the above algorithm showed excellent parity between the rotated-corrected transformations of the sample signature and new input signature from the same user, when the rotation angle varied between -30 to +30 degrees. The Rotation algorithm is in section 3.