Survey on Person Identification Using Soft Biometrics Arfa Mehek MounikaRaksha Reshma Banu Abstract

Survey on Person Identification Using Soft Biometrics
Arfa Mehek MounikaRaksha Reshma Banu
Biometrics is to do the work of Identification of humans by their own characteristics. The focus has been changed to multi-biometrics due to the security demands. Biometrics is divided into Hard Biometrics and Soft Biometrics. Hard Biometrics deals with the Physical, Behavioural and Biological characteristics such as facial features, eye, signature, voice, DNA and finger prints. Hard Biometrics uniquely identifies an individual. Whereas Soft Biometrics provides ancillary information that are not fully distinctive and permanent, so these features cannot provide a reliable person recognition. Soft Biometrics traits such as skin colour, skin texture, tattoos, shoe size, ethnicity, hair style, height, and weight have been regularly used as a secondary information to compliment the primary(hard) biometrics.

Keywords: Multi-biometrics, Hard biometrics, Soft biometrics
Soft biometric traits such as skin colour, tattoos, shoe size, height, and weight have been regularly used for forensic investigation, especially when hard biometric traits, e.g. face and finger prints are not available. With the increase in crime, surveillance systems have gained significant attention in the past few decades. In application domains such as surveillance scenarios in which Hard biometric traits may face challenges of different types of degradation, the use of soft biometrics has been shown to improve the performance attained with hard biometrics. In vision surveillance/biometric research, the development of system to work in unconstrained data acquisition protocols and uncontrolled lighting environments is the major ambition. These challenges arise from either environmental factor are due to lack of co-operation from user for instance, environmental challenges include illumination variation, background noise and blurriness. The soft biometric information that is extracted from a human body is easy to distinguish at a distance, but it is not fully distinguished and it can be identified in recognition tasks. The face recognition system is used to identify the person from the soft biometrics information. The description of facial hair and hair styles are among the most effectives soft biometrics traits is reported in this literature. Person head images, hair segmentations based on hair styles, moustaches and eyebrows, face shape, etc are some of the traits used in identification of individual. we study the discrimination capabilities of soft biometrics standalone. Then experiments are carried out fusing soft biometrics with two state-of-the-art face recognition systems. Soft biometrics are not expensive to compute since they can be acquired at the same time during hard biometric collection. Enables to generate a large set of hypotheses without compromising the time cost. Improvement of unconstrained facial recognition systems i.e, the degraded images are refined using different stages such as pre-processing, extraction and classification.

Pre-processing images is a method to convert an image into digital form and perform some operations on it in order to get an enhanced image or to extract some useful information from it. Feature Extraction starts from an initial set of measured data and builds derived values intended to be informative and non-redundant, which completely describes the original data set. Image Classification analysis the numerical properties of various image features and organises data into categories. Classification algorithms typically employ two faces of processing: training and testing. In the initial Training phase, characteristic properties of typical image features are isolated and, based on these, a unique description of each classification category, i.e. training classes, is created. In the subsequent testing phase, the feature-space partitions are used to classify image features.

Existing System:
Ester Gonzalez-Sosa, Julian Fierrez, Ruben Vera-Rodriguez, Fernando Alonso-Fernandez, “Facial Soft Biometric for Recognition in the Wild”, IEEE Transaction on information forensics and security, vol.13, no.8, august 2018
This paper employs the SFFS algorithm, we learn that soft biometrics are ranked from the most discriminative to the least in the following order: age, ethnicity, gender, moustache, glasses, beard. Then, the fusion of the face verification systems is analyzed along with a set of automatically estimated soft biometrics.

Hugo Proenca, Joao C Neves “Soft Biometrics: Globally coherent solutions for hair segmentation and style recognition based on hierarchical MRFs”, IEEE Transaction on Information Forensics and Security 12 (7), 1637-1645, 2016
The proposed MRF architecture is particularly suitable for problems that deal with biological data, where the reasonability of the solutions can be objectively measured. Markov Random Fields are a classical tool for many computer vision problems, from image segmentation, image registration to object recognition.

Shih-Hsiung Lee, Chu-Sing Yang, “An International Hair and Scalp analysis system”, International journal of Innovative Computing, Information and Control Volume 14, no.2, April 2018
This paper proposes the concept of using webcam and microscope camera sensors to extract characteristics images to evaluate the hair and scalp status of the user. Recently, owing to the enhancement in computation ability of intelligent devices and decrease in prices, it is feasible to have an inexpensive hair and scalp analysis system.
Askarsh Malhotra, Richa Singh, Mayank Vasta, Vishal M Patel, “Person Authentication using Head Images”, IEEE Winter Conference on Applications of Computer Vision (WACV),409-417,2018
In many surveillance applications, the cameras are placed at overhead heights for human identifications. In such real-world scenarios, the person of interest might be walking away from the camera and the only information available is image of the person’s head. Here we investigate the usage of head images for person recognition and propose it as a soft-biometric modality.
soft biometrics are either used as uni-modal systems, classification of traits such as single trait classifiers or the combination of other systems. We will discuss main domains of applications of soft biometrics.

Fusion with classical biometric traits: soft biometrics are made used in multi modal biometric systems in the aim of increasing the overall reliability, where the usage of soft biometrics to fingerprint had led to improvement of approximately over hard biometrics system.

Pruning the search: soft biometrics are used to filter large databases with the goal of increasing the efficiency. Research on soft biometrics for pruning search where the usage of multiple attributes such as age, skin colour is used for face identification. These attributes were used to enhance the performance of biometric system.

Human re-identification: for human re-identification the soft biometrics traits are related to limitations of accuracy, distinct result and performance are overcome by using multiple attributes altogether. Many works are having been done under the context mining and image retrieving. Many other applications are related to identify people based on their characteristics using statistical properties, facial characteristics and the samples of audio and others.