Off‐line signature verification and identification by pyramid histogram of oriented gradients

The purpose of this paper is to propose an effective method to perform off‐line signature verification and identification by applying a local shape descriptor pyramid histogram of oriented gradients (PHOGs), which represents local shape of an image by a histogram of edge orientations computed for each image sub‐region, quantized into a number of bins. Each bin in the PHOG histogram represents the number of edges that have orientations within a certain angular range.

Automatic signature verification and identification are then studied in the general binary and multi‐class pattern classification framework, with five different common applied classifiers thoroughly compared.

Simulation experiments show that PHOG has obvious advantages in the extraction of discriminating information from handwriting signature images compared with many previously proposed signature feature extraction approaches. The experiments also demonstrate that several classifiers, including k‐nearest neighbour, multiple layer perceptron and support vector machine (SVM) can all give very satisfactory performance with regard to false acceptance rate (FAR) and false rejection rate (FRR). On a public benchmarking signature database “Grupo de Procesado Digital de Senales” (GPDS), experiments demonstrate an FRR of 4.0 percent and an FAR 3.25 percent from SVM for skillful forgery, which compares sharply with the latest published results of FRR 16.4 percent and FAR 14.2 percent on the same dataset. Experiments on a second DAVAB off‐line signature database also illustrate the superiority of the proposed method. The related issue, off‐line signature recognition, which is to find the identification of the signature owner from a given signature database, is also investigated based on the PHOG features, showing superb classification accuracies of 99 and 96 percent for GPDS and DAVAB datasets, respectively.

The proposed method for off‐line signature verification and recognition has a promising potential of designing a real‐world system for many applications, particularly in forensics and biometrics.