Elephant herding with whale optimization enabled ORB features and CNN for Iris recognition

An Improved ORB Feature Extraction Algorithm Based on Enhanced Image and Truncated Adaptive Threshold

“…To address the problem of the unsuitability of the threshold in the ORB algorithm for different image scenes, scholars have proposed adaptive threshold calculation methods based on local pixel distribution, such as methods based on adaptive truncation threshold [22], adaptive median [23] and adaptive statistics [24]. In addition, machine learning methods, such as SVM (Support vector machine, SVM) [25], CNN (Convolutional neural networks, CNN) [26], have also been proposed to predict adaptive thresholds, thereby improving the robustness and adaptability of the ORB algorithm to complex lighting conditions.…”

Anti-Software Attack Ear Identification System Using Deep Feature Learning and Blockchain Protection

“…Ear recognition is a biometric technology that has emerged in recent years since ears, similar to fingerprints [1], irises [2], and faces [3], contain many specific and unique features [4] that can be used to identify a person [5]. The process of ear image acquisition is not dependent on the subject's cooperation and is non-invasive and non-contact.…”

“…Early ear recognition systems used hand-crafted features for recognition. These systems had four main drawbacks: (1) most did not use a baseline ear database for system performance testing; (2) no standard performance evaluation metrics were used; (3) the ear database used for recognition was obtained in a constrained environment, resulting in poor system generalization and poor performance on unconstrained ear databases; and (4) recognition performance lags behind that of systems based on deep feature learning. Recent years have witnessed the development of deep learning [11][12][13], and the technique has been used in ear recognition.…”

“…It can generate drop masks with adaptive shapes, which effectively improves the generally poor recognition performance of existing ear recognition systems when facing unconstrained ear databases in realistic scenarios. (2) We introduce a simple and parameter-free attention module [19] to infer the 3D attention weights of the ear feature maps output from the convolutional layer, enabling the ear recognition system to learn more critical feature regions. (3) We use Merkle tree [20] nodes to store the ear feature templates, ensuring the determinism of the root node in the smart contract and protecting the security of the ear feature template database and the user's privacy.…”