Method suggestion to verify handwritten signatures on limited surfaces

Today; Signing on paper is used as an identity indicator in official documents. In addition, the practice of signing documents produced on digital media such as tablets via the system has become widespread in recent years. In case of legal dispute; Verification of signatures on paper can be done by experts. On the other hand, in cases where signatures signed in digital environments may be fake, verifying the signatures claimed to have been made and their relationship with the document at hand requires complex methods. For this purpose, unlike the existing signature sample data sets in the literature, a new data set called Tatlıgil was created with the signatures donated by signature donors and a method proposal was presented to verify the signature. For this set, 20, 50 and 100 signatures belonging to the same donor were collected, along with records of right/left handedness, male/female, education level and age. In addition, the positions of the donors while signing were ensured and that they donated without interruption. It is evaluated that different studies to be conducted with the data set in the future will ensure that the accuracy and performance rate will be much higher and will add added value with different studies such as psychological status. The differences between our new data set and the existing ones are discussed. Studies with relevant data sets were examined and the method with the highest performance rate was determined. Its accuracy was compared with data sets in the relevant study. The high-performance method is related to the detection of forged signatures added between real signatures. However, when a person signs, his signature may change or deteriorate over time, when he signs repeatedly or due to his position while signing. For this reason, the method with high performance was tested on the data set we created. The method provided low performance and accuracy even though all signatures belonged to their real owners. It is considered that the reason for this low performance is that the signatures differ for various reasons, in other words, they are distorted. For this reason, if signatures are taken with a digital material, a method recommendation is presented regarding what should be taken into consideration when obtaining these signatures and what should be done to ensure verification.