Figure 1 - uploaded by Christoph Busch
Content may be subject to copyright.
Characteristics of the human ear the German criminal police uses for personal identification of suspects
Source publication
![Figure 2: Example images from the WPUT ear database [20]. The database...](profile/Christoph-Busch-2/publication/260664486/figure/fig3/AS:667690347794434@1536201192281/Example-images-from-the-WPUT-ear-database-20-The-database-contains-ear-photographs-of_Q320.jpg)
![Figure 3: Example images fron IIT Delhi ear database [21].](profile/Christoph-Busch-2/publication/260664486/figure/fig4/AS:667690347814916@1536201192351/Example-images-fron-IIT-Delhi-ear-database-21_Q320.jpg)
The possibility of identifying people by the shape of their outer ear was first discovered by the French criminologist Bertillon, and refined by the American police officer Iannarelli, who proposed a first ear recognition system based on only seven features. The detailed structure of the ear is not only unique, but also permanent, as the appearance...
Contexts in source publication
Context 1
... there is an ever-growing need to automatically authenticate individuals, biometrics has been an active field of research over the course of the last decade. Traditional means of automatic recognition, such as passwords or ID cards, can be stolen, faked, or forgot- ten. Biometric characteristics, on the other hand, are universal, unique, permanent, and measurable. The characteristic appearance of the human outer ear (or pinna) is formed by the outer helix, the antihelix, the lobe, the tragus, the antitragus, and the concha (see Figure ...
Context 2
... the skin elasticity. A first study on the effect of short periods of time on ear recognition [9] shows that the recognition rate is not affected by ageing. It must, however, be mentioned that the largest time elapsing difference in this experiment was only 10 months, and it therefore is still subject to further research whether time has a critical effect on biometric ear recognition systems or not. The ear can easily be captured from a distance, even if the subject is not fully cooper- ative. This makes ear recognition especially interesting for smart surveillance tasks and for forensic image analysis. Nowadays the observation of characteristics is a standard technique in forensic investigation and has been used as evidence in hundreds of cases. The strength of this evidence has, however, also been called into question by courts in the Netherlands [10]. In order to study the strength of ear prints as evidence, the Forensic Ear identification Project (FearID) was initiated by nine institutes from Italy, the UK, and the Netherlands in 2006. In their test system, they measured an Equal Error Rate (EER) of 4% and came to the conclusion that ear prints can be used as evidence in a semi-automated system [11]. The German criminal police use the physical properties of the ear in connection with other appearance-based properties to collect evidence for the identity of suspects from surveillance camera images. Figure 1 illustrates the most important elements and landmarks of the outer ear, which are used by the German BKA for manual identification of suspects. In this work we extend existing surveys on ear biometrics, such as [12], [13], [14], [15] or [16]. Abaza et al. [17] contributed an excellent survey on ear recognition in March 2010. Their work covers the history of ear biometrics, a selection of available databases and a review of 2D and 3D ear recognition systems. This work amends the survey by Abaza et al. with the ...
Similar publications
The outer ear features have been used for many years in forensic science of recognition. Human ear is a valuable information provenance of data for individual identification/authentication. Ear meets biometric characteristic (universality, distinctiveness, permanence and collectability). Biometric system depending on ear image facing two major chal...
Citations
... As a result, the ear can be a reliable biometric identification since it has the characteristic known as uniqueness [9]. Another advantage is that it does not require a person's cooperation for this biometric to be collected [21] and consequently used to identify an individual. Therefore, situations involving surveillance and tracking routines benefit from such biometrics [15]. ...
... The determination and recreation of its individual properties are important for applications in diverse fields. For example, the pinna shape has been used to identify a person [1,2] and to reconstruct a subject's morphology after accidents or medical problems with their pinna [3]. Moreover, the details of the individual pinna shape affect the acoustic filtering of sound arriving at the ear canal, typically described by the so-called head-related transfer functions (HRTFs) [4]. ...
... A shape key is a tool in Blender that implements a deformation of mesh parts with the degree of deformation being described by a scalar parameter. Conceptually, a shape key realizes linear interpolations between two limit meshes (0) and (1) , which usually only differ in a region, i.e., by the positions of vertices in a local (connected) group. 4 In BezierPPM, we consider 17 shape keys, which are listed in Table 1. ...
A parametric representation of the complex biological structure of the human pinna is of considerable interest in applications such as the design of ear prostheses, personalization of spatial hearing, and biometric identification. Here, we describe BezierPPM, a parametric pinna model with parameters closely linked to human ear structures. BezierPPM represents the ear geometry as cubic Beziér curves and includes local modifiers of predefined concave areas. We evaluated BezierPPM by manually registering its parameters to 20 ears selected from various databases of digitized ears. The root-mean-square error between the registered and target geometries was on average 1.5 ± 0.2 mm, and the 2-mm and 1-mm completeness was 87 ± 4% and 66 ± 5%, respectively. This indicates that BezierPPM was capable of representing human pinnae very well. An in-depth analysis showed that most of the inaccuracies were located on the back side of the ear, an area having a rather low relevance for most target applications. To address potential future applications in machine-learning settings, we used BezierPPM to create a database of synthetic pinna geometries and trained a toy-example neural network to estimate BezierPPM parameters from multi-view images of this database. The estimated parameters resulted in estimated pinna geometries with a mean error of 0.3 mm, indicating that BezierPPM can accurately describe a wide variety of human pinnae, also in machine-learning settings. In summary, our evaluations demonstrate a high potential of BezierPPM for applications requiring a good representation of the relevant parts of pinna geometry, especially when targeting applications in machine-learning settings.
... Choudhary et al. [15] analysed various feature extraction methods for iris recognition and compared the performance of different feature extraction methods. Pflug and Busch [16] surveyed the algorithms of ear detection and recognition using both 2D and 3D images. Many feature extraction methods are introduced and discussed. ...
Biometric recognition is a widely used technology for user authentication. In the application of this technology, biometric security and recognition accuracy are two important issues that should be considered. In terms of biometric security, cancellable biometrics is an effective technique for protecting biometric data. Regarding recognition accuracy, feature representation plays a significant role in the performance and reliability of cancellable biometric systems. How to design good feature representations for cancellable biometrics is a challenging topic that has attracted a great deal of attention from the computer vision community, especially from researchers of cancellable biometrics. Feature extraction and learning in cancellable biometrics is to find suitable feature representations with a view to achieving satisfactory recognition performance, while the privacy of biometric data is protected. This survey informs the progress, trend and challenges of feature extraction and learning for cancellable biometrics, thus shedding light on the latest developments and future research of this area.
... We have proposed Hessian Matrix Determinant as an algorithm for property detection. There are many detection methods for 2D and 3D images that have been summarized in [22], however, any type of texture-based detection method has not been investigated. ...
... The geometric approach utilizes statistics-based geometric properties of the ear images. These methods are invariant to geometric distortion like scaling, rotation, and other small changes (Pflug and Busch 2012). Hurley et al. proposed a force field-based feature extraction technique that utilizes holistic property. ...
... Performances of different techniques are evaluated using Rank-1(R1) accuracy and Equal Error Rate (EER). Deep learning-based methods perform the classification task based on the relevant features with the help of Convolution and MaxPooling layers (Xie and Mu 2008;Emeršič et al. 2017a;Pflug and Busch 2012). ...
Person identifications using the ear-based biometric system has become quite popular in recent year due to increasing demands in security and surveillance applications. With limited training data and computing resources, the run time complexity plays an important role in such a biometric system. With the continuous advancement in deep convolutional neural networks, deep learning-based biometric systems consequently achieved huge progress in solving earlier unanswered and/or incomplete challenges. Though ear-based biometric system gives higher accuracy with the help of pre-trained deep learning models like VGG19, VGG16, Xception, etc. Training these models is a cumbersome task and it requires much time. Most of the ear recognition system developed using deep learning models like VGG19, Xception, ResNet101, etc. requires a large memory area due to the huge parameter requirement of the model. Also, they put computational overhead on the system. One of the major challenges in the field of ear recognition is to identify people with the help of electronic devices over time and space. While developing electronic approaches for person identification, it is worth important to consider the factors like simplicity, cost-effectiveness, and portable flexibility. With these motivations, the authors developed three simple lightweight CNN models and ensemble them to get improved recognition accuracy. The work is validated on the IITD-II ear dataset which contains only 793 sample mages for training purposes. To overcome the limitation of the limited dataset, the author performed data augmentation technique which produces a variety of images from different perspectives. By stacking these three CNN models, an optimal architecture is developed that gives the best accuracy of 98.74% which is a good improvement over the individual model. The proposed CNN models can also be ensemble with other pre-trained models like VGG16, VGG19, ResNet, Xception, etc. for a more effective solution.
... With the progressive developments of ear recognition system, a number of reviews and survey exist that summarizes the challenges and limitation of ear based recognition system [6][7][8][9]. Presence of accessories and high degree of head movement that significantly impact the recognition performance are reported by the authors [10]. ...
... Geometric approach deals with the geometric properties of the ear images which are mainly statistics based. These methods are invariant to geometric distortion like scaling, rotation and other small changes [7]. Fourier transform-based methods have been reported in literature and it is also useful to design rotation invariant descriptor to extract the ear features [27]. ...
... These methods are mainly based on keypoints [13] detection and descriptions. Among the local approaches [6,[38][39][40] scale invariant feature transform (SIFT) [7,41,42], histogram of oriented gradients (HOG) [43], speed up robust features (SURF) [44], Local Binary Pattern (LBP) [45] and Log-Gabor Filters [46] are mainly used for detecting the key points to extract the texture information over the entire image [38,39,40]. The geometric, holistic and local approaches combine and give another approach called hybrid approach. ...
In the pandemic of COVID-19, identifying a person from their face became difficult due to wearing of mask. In regard to the given circumstances, the authors have remarkably put effort on identifying a person using 2D ear images based on deep convolutional neural network (CNNs). They investigated the challenges of limited data and varying environmental conditions in this regards. To deal with such challenges, the authors developed an augmentation-based light-weight CNN model using CPU enabled machine so that it can be ported into embedded devices. While applying data augmentation technique to enhance the quality and size of training dataset, the authors analysed and discussed the different augmentation parameters (rotation, flipping, zooming, and fill mode) that are effective for generating the large number of sample images of different variability. The model works well on constrained and unconstrained ear datasets and achieves good recognition accuracy. It also reduces the problem of overfitting.
... At present, many kinds of biological characteristics can be used for personal identification and they can be mainly divided into two categories: physiological characteristics and behavioural characteristics. Common physiological characteristics include human faces [2], fingerprints [3], hands [4,5], irises [6,7], and ears [8,9]; behavioural characteristics include voices [10], gaits [11], signatures [12,13], and keystrokes [14,15]. However, these human biological characteristics are easy to falsify, which makes this type of approach to personal identification vulnerable. ...
... The first sur-vey paper was published by Pun and Moon [8] (2004) to the best of our knowledge. Later, other survey papers were published; the most famous are: Choras [9] (2007), Ramesh and Rao [10] (2009), Kurniawan et al. [11] (2012), Pflug and Busch [12] (2012), Abasa et al. [13] (2013), Emeršič et al. [14] (2017), Alva et al. [15] (2019), Srivastava et al. [16] (2020), Wang et al. [17] (2021), Kamboj et al. [18] (2021), etc. In Table 1, we summarized some specific points and aspects to differentiate and compare the most famous ear recognition surveys and highlight our main contributions. ...
... We can observe that Pflug and Busch [12] (2012) and Abasa et al. [13] (2013) presented comprehensive and rich survey papers that consider detailed sections dealing with ear detection, datasets, 2D and 3D ear recognition approaches. We should also recall the interest of the surveys [12,13] which correspond to the foundations of ear recognition. ...
... We can observe that Pflug and Busch [12] (2012) and Abasa et al. [13] (2013) presented comprehensive and rich survey papers that consider detailed sections dealing with ear detection, datasets, 2D and 3D ear recognition approaches. We should also recall the interest of the surveys [12,13] which correspond to the foundations of ear recognition. In our work, we complement them by relying on new concepts such as Deep Learning. ...
Automatic identity recognition from ear images is an active research topic in the biometric community. The ability to secretly acquire images of the ear remotely and the stability of the ear shape over time make this technology a promising alternative for surveillance, authentication, and forensic applications. In recent years, significant research has been conducted in this area. Nevertheless, challenges remain that limit the commercial use of this technology. Several phases of the ear recognition system have been studied in the literature, from ear detection, normalization, and feature extraction to classification. This paper reviews the most recent methods used to describe and classify biometric features of the ear. We propose a first taxonomy to group existing approaches to ear recognition, including 2D, 3D, and combined 2D and 3D methods, as well as an overview of historical advances in this field. It is well known that data and algorithms are the essential components in biometrics, particularly in-ear recognition. However, early ear recognition datasets were very limited and collected in laboratory with controlled environments. With the wider use of deep neural networks, a considerable amount of training data has become necessary if acceptable ear recognition performance is to be achieved. As a consequence, current ear recognition datasets have increased significantly in size. This paper gives an overview of the chronological evolution of ear recognition datasets and compares the performance of conventional vs. deep learning methods on several datasets. We proposed a second taxonomy to classify the existing databases, including 2D, 3D, and video ear datasets. Finally, some open challenges and trends are debated for future research.
... Ref. [31] conducted preliminary research on Force Field Transformation (FFT) for automatic ear recognition and returned a recognition rate of 99% on about 252 images in the XM2VTS database. Ref. [32] furthered the application of FFT with the underlying principle of Newton's law of gravitation to consider symmetric image pixels. ...
... Gabor filters are also capable of identifying detailed texture data. When fused, its recognition accuracy varies between 92.06% and 95.93% [32]. Dimensionality reduction techniques such as PCA [31,34], ICA [35] and matrix factorization [36], feed higher-dimension vectors into lower dimensions while retaining their distinct features. ...
... Studies such as [40] present SIFT as a robust algorithm suitable for feature extraction under changing conditions. For instance, SIFT can accommodate variants in the pose for about 20 degrees [32]. Generally, assigning landmarks to ear images before training ensures proper filtering and matching operations in the local technique. ...
Biometric technology is fast gaining pace as a veritable developmental tool. So far, biometric procedures have been predominantly used to ensure identity and ear recognition techniques continue to provide very robust research prospects. This paper proposes to identify and review present techniques for ear biometrics using certain parameters: machine learning methods, and procedures and provide directions for future research. Ten databases were accessed, including ACM, Wiley, IEEE, Springer, Emerald, Elsevier, Sage, MIT, Taylor & Francis, and Science Direct, and 1121 publications were retrieved. In order to obtain relevant materials, some articles were excused using certain criteria such as abstract eligibility, duplicity, and uncertainty (indeterminate method). As a result, 73 papers were selected for in-depth assessment and significance. A quantitative analysis was carried out on the identified works using search strategies: source, technique, datasets, status, and architecture. A Quantitative Analysis (QA) of feature extraction methods was carried out on the selected studies with a geometric approach indicating the highest value at 36%, followed by the local method at 27%. Several architectures, such as Convolutional Neural Network, restricted Boltzmann machine, auto-encoder, deep belief network, and other unspecified architectures, showed 38%, 28%, 21%, 5%, and 4%, respectively. Essentially, this survey also provides the various status of existing methods used in classifying related studies. A taxonomy of the current methodologies of ear recognition system was presented along with a publicly available occlussion and pose sensitive black ear image dataset of 970 images. The study concludes with the need for researchers to consider improvements in the speed and security of available feature extraction algorithms.
... While many powerful recognition approaches, mostly based on deep learning, have been proposed in the literature recently, the problem of ear alignment has received comparably less attention, as also emphasised in recent surveys in this field. [2,3], Despite its importance and (potentially beneficial) impact on all downstream tasks, efficient ear alignment in diverse settings is still largely unsolved. ...
Ear images have been shown to be a reliable modality for biometric recognition with desirable characteristics, such as high universality, distinctiveness, measurability and permanence. While a considerable amount of research has been directed towards ear recognition techniques, the problem of ear alignment is still under‐explored in the open literature. Nonetheless, accurate alignment of ear images, especially in unconstrained acquisition scenarios, where the ear appearance is expected to vary widely due to pose and view point variations, is critical for the performance of all downstream tasks, including ear recognition. Here, the authors address this problem and present a framework for ear alignment that relies on a two‐step procedure: (i) automatic landmark detection and (ii) fiducial point alignment. For the first (landmark detection) step, the authors implement and train a Two‐Stack Hourglass model (2‐SHGNet) capable of accurately predicting 55 landmarks on diverse ear images captured in uncontrolled conditions. For the second (alignment) step, the authors use the Random Sample Consensus (RANSAC) algorithm to align the estimated landmark/fiducial points with a pre‐defined ear shape (i.e. a collection of average ear landmark positions). The authors evaluate the proposed framework in comprehensive experiments on the AWEx and ITWE datasets and show that the 2‐SHGNet model leads to more accurate landmark predictions than competing state‐of‐the‐art models from the literature. Furthermore, the authors also demonstrate that the alignment step significantly improves recognition accuracy with ear images from unconstrained environments compared to unaligned imagery.
