By all is known the proportionality of two signatures, whether we sign in a small space or we do in a larger, the signature will not be exactly the same (If two signatures are exactly the same one of them is false), but will retain a general proportionality unique to the signatory.
Often the proportional analysis of the general geometry of the signature is of great help, this type of analysis has some advantages, for example if the image quality is bad, it will not matter, since, the analysis is done based on The actual dimensions of the strokes, not the color tones of the pixels.
We have tried to use a mathematical method, never used to analyze proportionality in writing, the golden ratio, and also to develop a revolutionary algorithm, which not only analyzes the real measures of proportion, but also takes into account other factors in the calculation, such as Example a percentage of confidence, which varies the final result, if the two firms are totally equal, since this denotes an axiomatic falsification.
To do this after several months of work and empirical tests, we present a new FDM (Fibonacci Digital Modeller) tool that will now incorporate NEGA, it will be free for users with the active license, since, they will be able to update to the new version .
The method used is based on the mathematical succession of Fibonacci, in this infinite sequence, there is a ratio between successive numbers that make up that tends to a defined limit, as we increase the series, the golden number, 1.618.
For this reason this series is used to define the golden geometry throughout the firm, generating the resulting golden spiral, a special algorithm has been devised that automatically searches for and positions points around the perimeter of the signature, following the golden quadrants of the Spiral generated, these points can be manually positioned, with the mouse, in case the algorithm has considered any line, part of a rubber stamp or buffer, an anfirm, a line or something else, mistaken for a stroke of the Signature, and this causes erroneous results, to do this has been designed a very easy color code, which indicates in which area of the quadrant, we have to position the point, so that it is within the trajectory of the stroke.
The calculation performed by the algorithm is based on a proportional system, so it is not affected by the size of the signature photo.
The FDM analysis is done from photos classified by project, previously saved and generated in NEGA of the signatures that we want to analyze. They are then analyzed in the FDM tool one by one and the analysis is saved. Subsequently multiple collations can be generated for each project, and generate a graph complementary to the final percentage results. The analyzes are done by doubles automatically, always combining the indubitada signatures (Know) with all the doubts (Questioned) that are in the list. The final result is presented in HTML format, which can be copied from the screen with the mouse and easily pasted into a Word-type text editor or an Excel spreadsheet. All project files are saved, if you want to consult later, in a directory on the desktop, called, “My Projects NEGA_ACPC”.
Due to the complexity of the mathematical analysis, we have chosen to offer three percentage results.
Accuracy of collation
It indicates a real percentage of proportional accuracy, which is reflected in the graph, similar graphs will have a comparison accuracy of 100%, but that does not mean that the writings are not false, because two signatures are false, a person does not Signature always the same in terms of measures, but retains proportionality in the general form of the signature.
Confidence of the collation
This value will indicate the level of confidence of the previous section, if the accuracy of the comparison is too high, we will have a low value in this section, it will stabilize, to take higher values, in checking accuracies that are acceptable, For accuracy values below 80%.
It is the final result, which presents a percentage of Fibonacci similarity indicating the following background colors:
If the result is between 80% and 100% (positive similarity)
If the result is between 60% and 79% (indeterminate similarity)
If the result is below 60% (negative similarity)
This value is calculated taking into account the previous data of accuracy and confidence.