Handwritten signature recognition using Fraunhofer diffraction pattern

Pattern recognition is a subfield of artificial intelligence and machine learning that often requires large datasets to identify the desired patterns. Consequently, the training process can be time-consuming, and recognizing certain objects may take several hours of computation. By contrast, Fraunhofer diffraction patterns of objects can be obtained quickly and easily through numerical methods, as these patterns are mathematically equivalent to a Fourier transform. We propose a method for recognizing handwritten signatures using their Fraunhofer diffraction patterns. The method does not require prior training with large volumes of data. Instead, it relies on the direct correlation of the diffraction patterns obtained for each signature, which makes the process fast and efficient. Notably, this approach is capable of identifying a set of 645 handwritten signatures in 19 s with an accuracy of 95.66%. To further validate its effectiveness, we quantitatively compared our method with a trained support vector machine (SVM) classifier using the same feature representations. The proposed approach significantly outperformed the SVM in both recognition accuracy and processing time. The limitations of the method are illustrated by applying it to the recognition of handwritten letters.