This method generates string for the given original signature. Basically by using this method each image gets converted into a string. So the original signature which is input to the method, a string can be constructed. Now if we rotate this signature by 0 to 360 degrees then we will get 360 different signatures and we can perform translation and scaling operation on each of the 360 signatures. So we will get more numbers of signatures. This method will collect the signature to be recognized as input and will construct the string. The generated signature now will be matched with the string of the original signature. If there is no match then the string will be matched with the strings already constructed or dynamically constructed so that memory requirement will be less. The input to the method signature will be compared with the signature which are allowed in the domain( certain degrees of rotation is practically allowed , by certain pixels the image can be practically shifted or translated).If it matches with any of the string pattern then it is recognized. Certain error levels can also be considered like number of pixels in the rows, relative differences in the positions of the pixels.
[...] (Containing 1598 pixels) String for the First transformation (Containing 1598 pixels) String for the Second transformation (Containing 1598 pixels) String for the Third transformation (Containing 1598 pixels) RELATED WORK Relative differences of the strings with respect to the original string are shown below For the First Transformation (Containing 1598 pixels) For the Second Transformation (Containing 1598 pixels) For the Third Transformation (Containing 1598 pixels) Then signature is rotated and new image is constructed by using the standard MATLAB function and I got 1550 pixels in the rotated image. [...]
[...] AP: - Applied Signature (Sample signature of the same person at different time) TP: - Total Pixels (Total number of active pixels of the original signature) MP: - Matched Pixels PM: -Percentage Match Pixels Missing with Absolute Difference of one Pixels Missing with Absolute Difference of Two Pixels Missing with Absolute Difference of Three Pixels Missing with Absolute Difference of Four Pixels Missing with Absolute Difference of Five TABLE 1 ANALYSIS OF SD.JPG SIGNATURE AP TP MP PM D1 D2 D3 D4 D5 AP TP TABLE 3 ANALYSIS OF AT.JPG SIGNATURE MP PM D1 D2 D3 D4 D5 sd1 sd2 sd3 sd4 sd5 sd6 sd7 sd8 sd at1 at2 at3 at4 at5 at6 at7 at8 at If observe carefully above table we find that sd.jpg image contains total 2256 active pixels and out of these 1934 are exactly matched( for sd1.jpg which is applied for the recognition ) , so the percentage match is 86. [...]
[...] The string generated is the unique string for the image by extracting all the features of the signature. Structure of the string will remain unique after performing all the transformation. Consider one sample example of the image shown below. The image I have considered is the black and white image so only two intensity levels are considered. Pixel with value 0 is the black pixel and the pixel with the value 1 is the white pixel. Now after performing translation operation in the x direction by I got the string.The string is like this Now after performing translation operation in the y direction by I got the following string is the output string.Now after performing translation operation in the y direction by 1 and x direction by I got the string is the generated string. [...]
[...] Below mentioned is the algorithm for the matching of the string created for the given input signature with the strings for all the allowable transformation. Repeat following steps for all the strings generated for all the allowable Transformations Algorithm Row and the column number can be determined for all the active pixels of the signature. Total numbers of the pixels per row can be determined and can be matched with the input string. Allowable Error can be set for the rows. [...]
[...] Initially a string is constructed for the rotated image and then I have applied the transformations and constructed the strings and found out the relative differences of the generated strings with respect to the original string (Rotated Image).The original image is rotated in the anti clockwise direction by two degree and corresponding strings are constructed. Again I have applied the same transformations those I have applied on the original image and found out the four strings. Here after applying the rotation transformation we will get a string and comparison operation now will be performed with the input applied signature and the rotated signature. [...]
using our reader.