Koşmaz Sünnetci, Betül; Bingöl, Özkan; Gedikli, Eyüp; Ekinci, Murat; Doğan, Ramazan Özgür; Türk, Salih; Güngör, Nihan

Featured Application: The proposed framework facilitates secure and user-friendly biometric authentication on mobile platforms, particularly in mobile payment and access control scenarios, under unconstrained conditions. The widespread adoption of mobile devices has made secure and user-friendly biometric authentication critical. However, widely used modalities such as fingerprint and facial recognition show limited robustness under uncontrolled illumination and on heterogeneous devices. In contrast, palmprint recognition offers strong potential because of its rich textural patterns and high discriminative power. This study addresses the limitations of laboratory-based datasets that fail to capture real-world challenges. We introduce MPW-180, a novel dataset comprising videos of 180 participants recorded on their own smartphones in everyday environments. By systematically incorporating diverse illumination conditions (with and without flash) and natural free-hand movements, MPW-180 is the first dataset to adopt a bring-your-own-device paradigm, providing a realistic benchmark for evaluating generalization in mobile biometric models. In addition, we propose PalmWildNet, an SE-block-enhanced deep learning architecture trained with Triplet Loss and a cross-illumination sampling strategy. The experimental results show that conventional methods suffer over 50% performance degradation under cross-illumination conditions. In contrast, our method reduces the Equal Error Rate to 1–2% while maintaining an accuracy above 97%. These findings demonstrate that the proposed framework not only tolerates illumination variability but also learns robust illumination-invariant representations, making it well-suited for mobile biometric authentication.

Sünnetci Koşmaz, Betül; Bingöl, Özkan; Gedikli, Eyüp; Ekinci, Murat

In palm biometric recognition systems, the hand being at different angles and positions relative to the camera presents a critical challenge in terms of accuracy and reliability. Especially in images captured in free environments, hand misalignment can negatively affect recognition performance by making it difficult to extract biometric attributes. Perspective distortions due to camera position change the geometric structure of the hand, compromising biometric data integrity. While 3D imaging systems can offer successful solutions for pose normalization, performing a similar transformation for 2D images poses a major technical challenge. In this study, depth estimation supported plane fitting is used to remove perspective distortions in 2D images and pose normalization of the hand is obtained. The proposed method uses the information obtained from depth estimation to realign the hand image as if it were standing directly in front of the camera, improving geometric consistency. This will make the extraction of biometric features more reliable and contribute to improving the accuracy of palm recognition systems.

MacAn, Huseyin Furkan; Bingol, Ozkan; Turk, Salih; Ekinci, Murat

The sizes of regions of interest (ROI) obtained from images captured by cameras in palmprint recognition may vary depending on the camera specifications and the position of the hand. The regions of interest that recognition systems can use are typically rescaled through interpolation techniques and then defined as inputs to the system. In such cases, there may be some loss of biometric data. This study aims to enhance the quality of the palmprint ROI images, which are critical for palmprint recognition systems. To achieve this, various super-resolution techniques were employed to reconstruct low-resolution ROI images to the target sizes while preserving the distinctiveness of their biometric features. This approach is designed to optimize recognition accuracy and improve data reliability in palmprint recognition processes.

Kosmaz Sunnetci, Betul; Bingol, Ozkan; Gedikli, Eyup; Ekinci, Murat

Automatic ROI extraction from hand images is a critical step in palmprint recognition. Traditional methods involve steps such as identifying the region of the hand, finding the valleys between the fingers and marking the palm area. Some of these processes are performed manually and some are performed with image processing techniques. However, these methods are time-consuming, vulnerable to human error, and computationally redundant. In this study, a deep learning-based convolutional neural network is developed for high-performance automatic ROI extraction from hand images. The proposed method eliminates the hand segmentation process and provides a solution where the palm region can be automatically selected directly. This reduces processing time, minimizes human error and improves overall accuracy.

Akpulat, Murat; Ekinci, Murat

The segmentation of moving and non-moving regions in an image within the field of crowd analysis is a crucial process in terms of understanding crowd behavior. In many studies, similar movements were segmented according to the location, adjacency to each other, direction, and average speed. However, these segments may not in turn indicate the same types of behavior in each region. The purpose of this study is to better understand crowd behavior by locally measuring the degree of interaction/complexity within the segment. For this purpose, the flow of motion in the image is primarily represented as a series of trajectories. The image is divided into hexagonal cells and the finite time braid entropy (FTBE) values are calculated according to the different projection angles of each cell. These values depend on the complexity of the spiral structure that the trajectories generated throughout the movement and show the degree of interaction among pedestrians. In this study, behaviors of different complexities determined in segments are pictured as similar movements on the whole. This study has been tested on 49 different video sequences from the UCF and CUHK databases.

Akpulat, Murat; Ekinci, Murat

The segmentation of the moving and non-moving parts in an image within the crowd analysis studies is a crucial result in terms of understanding the behaviour. However, in many studies carried out, moving areas are assessed as making a set of movement that are similar in terms of location and of acceleration. In this study, what kind of interaction there exists among the people in a high-density crowd movement was studied. Initially, the motion data is represented as trajectories by using the optical flow and particle advection methods. The image is divided into equal hexagons and each area's topological entropy value is determined through Finite Time Braid Entropy (FTBE). This value depends on the complexity of the spiral structure that the trajectories generate throughout the movement and this value shows how much effective it is in the area. Eventually, through the complexity map that was generated, the behaviour of the crowd could be better understood and assessed. Also, in this way, possible abnormal conditions in the area can be predicted by monitoring each region and by setting off from the changes in the values of Braid entropy in time. The method was tested using the high-density crowd movements in the UCF database and results were presented.

Akpulat, Murat; Ekinci, Murat

Providing security and managing it as well as trying to understand how it behaves have become an important research topic in our day. The aim of this study is to calculate the entropy value of the crowd movement regionally and to track how it. changes in time. Because, excessive changes in the entropy value in time might point to abnormal behaviour in the region. The motion information obtained from the optical flow method from the image in parallel to the target was converted to motion orbits by particle advection. Motion orbits are mapped into by the Braid Theory and entropy value is calculated. Braid Theory is a mathematical method that is used to express entropy as a qualitative value by determining the interaction among the movement orbits. The results are tested according to the video sequence in the PETS2009 and UCF database and they show that this method is functional for observing how the entropy changes in time within the crowd movement.

Bingöl, Özkan; Ekinci, Murat

In order to increase performance in palmprint recognition systems, various devices are normally used to restrict the movement of the hand. These can cause problems, especially for those users with physical disabilities. They also cause significant hygiene problems in multi-user systems. Recently, studies on palmprint recognition systems have progressed towards the development of unconstrained, contactless and unrestricted background techniques. The most common problem encountered in these studies is the alignment arising from the free movement of the hand. Despite 3D hand-acquisition devices which offer extra recognition features to overcome this problem, the applicability of these devices is low because of their increased cost. In this study, a stereo camera was proposed. Although due to matching problems, it is difficult to achieve precise, distinct feature extraction in the unrestricted 3D environment used for palmprint recognition, the orientation of the hand in 3D space can be determined by obtaining depth information. In this study, the depth information was extracted by using the binocular stereo approach. First, the orientation of the hand was estimated by fitting a surface model associated with the eigenvectors of the depth information. Pose correction was then accomplished by establishing a relationship between the orientation and the images. The pose correction greatly relieved the perspective distortion that usually occurs within the various poses of the hands. Next, the region of interest was determined by performing segmentation on the corrected images using the Active Appearance Model (AAM). The palmprint features were then extracted via Gabor-based Kernel Fisher Discriminant Analysis. In order to demonstrate the performance of the proposed approach, a new dataset was compiled from stereo images within various scenarios collected from 138 different individuals. As a result of these experimental studies, the EER values, especially on the images captured from different hand orientations in 3D, were reduced from around 14–0.75%. With the help of this suggested approach, the palmprint recognition system was transformed into a more portable form by removing the closed-box mechanisms and equipment restricting movement of the hand. This system can automatically perform pose estimation, hand segmentation and recognition processes without any special intervention.

Akpulat, Murat; Ekinci, Murat

Today, trying to understand what kind of behaviour the crowd shows by studying the data from surveillance systems is an important topic for researchers of computer vision. The aim of this study make the motion data that is at pixel level and that is obtained by optical flow method a more meaningful data set with the particle advection method. In other words, the aim is to monitor the motion data by converting the 3D optical flow data set to a 2D data set. With this respect, evaluating how the crowd motion changes throughout the video by focusing only on the moments where there is motion can be carried out easily. At the end of the study, the success of the method is presented by testing how the image acts as a dominant movement.

Bingöl, Ozkan; Ekinci, Murat

In recent years, biometric palmprint recognition system is one of the preferred research areas due to reach high accuracy even with the low resolution cameras. In recent studies conducted in this area, especially due to hygiene problems, images are taken in contactless and unrestricted environment. Despite a very successful improving methods have been developed for the images obtained in the restricted environment, shifts or projective distortions that occur in the palm pattern obtained from images taken in unconstrained environments negatively affect the success of this method. In this study, coding techniques that are commonly used palmprint recognition systems (Competitive Code, Ordinal Code, Derivative of Gaussian Filters Code) are strengthened with feature determinants (SIFT, SURF) and a new approach has been proposed. Scope of the study, a new data set consists of low-resolution images taken in unconstrained environments created and palmprint patterns obtained with the help of Active Appearance Model. First, performance rate of these coding techniques on the patterns were analyzed. Then, the advantages of 6 different approaches obtained by combining each feature determinant with each coding technique, is disclosed to the system.

Bingol, Ozkan; Ekinci, Murat; Gokdogan, Ahmet

Regarding the palms recognition system studies, despite achieving a high success rate, hygiene problems in systems with contact and problems arising from changes in the alignment of the hand pose in non-contact ones have been encountered. To resolve these problems, 3D palmprint recognition systems have been developed, however these systems have not had the opportunity to spread due to expensive technologies used as well as low screening rates. In this study, to calculate the position and orientation of a palm in 3D space, a stereo camera system is proposed, with the help of these values by adjusting the geometric image of the palm is converted to a 2D environment. From these images, patterns used for recognition are extracted by using Active Appearance Models. Thus a non-contact system has been achieved, but also patterns that can be used by powerful 2D recognition systems have also been produced.

Bingöl, Ozkan; Ekinci, Murat

Although studies on the palm recognition systems achieving are high performance ratios, hygiene problems in contact systems and change of hand pose in contactless systems were encountered. 3D recognition systems is proposed for solve these problems. But, these systems did not find the opportunity to become widespread because of expensive technologies and low scanning speed. In this study was proposed a method for detemined pose variations of palmprint using stereo cameras and transferring patterns that to be used for recognition to 2D space. Thus, both a non-contact system will be obtained and can be used properties of strong 2D recognition systems.