The goal of signature verification systems is to ascertain whether or not a given signature is genuine (i.e., made by the individual who is being claimed as the signer) or whether it is a forgery (produced by an impostor). This is a difficult procedure in general, but it is especially challenging in an offline (static) case that involves scanned signature images. A handwritten signature has a significant advantage over other types of biometric technology, such as fingerprint or voice verification, due to the fact that it is the form of identity verification that is most frequently recognised as a biometric. Within the scope of this study, Offline HOG-based Signature Recognition is presented. This technique counts the occurrences of discrete regions within a picture by using the direction of gradients present in those regions. In order to accomplish this, we employ a very small grid that has very small cells for the computation of characteristics. The recognition algorithm that is employed is called KNN. During the course of the tests, a total of 60 photographs were used for evaluation, while the remaining 100 were used for training. We present evidence that demonstrates that the derived FAR for a K = 1 population of FDs and HOGs is 0.1857 and 0.0600, respectively. It was discovered that HOG had a higher level of recognition accuracy than FD features. Within the scope of this case study, we present the recognition results for ten participants, with each subject having been trained on ten photographs and having been tested on six images.