Aruni Singh

Newborn's ear recognition: Can it be done?

Abduction, swapping and mix ups are the unfortunate events that could happen to infants while in hospital premises and medical personnel are finding it difficult to curb this unfortunate incidents. Existing biometric and non biometric methods fail to provide enough level of security and research done to solve this problem is minimal. This paper introduces the concept of using ear recognition for identification of newborn and presents automatic ear recognition algorithm. The main contribution of this research is (i) preparation of a new born ear database of 210 individuals with slight variations in illumination and pose (ii) testing of different ear matching algorithms. On an ear biometric database of 210 infants it is concluded that ear can be used as biometric feature to identify the newborn.

Intelligent Method for Face Recognition of Infant

Mixing, abduction and illegal adoption of infant is a global challenge and the research done to solve this problem is minimal. Traditional methods of biometric and non biometric techniques have their own limitations and fail to provide required level of security. The task of face recognition has been actively researched in recent year and automatic face recognition of infant is not reported in the literature. The contributions of this research are: (i) Design and implementation of intelligent face recognition technique to recognize infant. (ii) Preparing an infant face database of 150 individuals with slight variations in pose, illumination; expression and quality. The results on the infant database are very encouraging.

Can face and soft-biometric traits assist in recognition of newborn?

Most of the biometric research being done is for adults and the identification accuracy of newborn are least reported in the literatures. In this paper we propose a novel biometric identification method for newborn babies using their face and soft biometrics. Accurate patient identification (ID) is essential for patient safety, especially with our smallest and most vulnerable pediatric patients. The main contribution of the research are (a) the preparation of face and soft biometric database of newborn. (b) Testing the algorithm for identification of 210 newborn. The combined use of all the four soft biometric traits results in an improvement of approximately 6% over the primary biometric system.

Performance of Face Recognition Algorithms on Dummy Faces

Face recognition is becoming increasingly important in the contexts of computer vision, neuroscience, psychology, surveillance, credit card fraud detection, pattern recognition, neural network, content based video processing, assistive devices for visual impaired, etc. Face is a strong biometric trait for identification and hence criminals always try to hide their face by different artificial means such as plastic surgery, disguise and dummy. The availability of a comprehensive face database is crucial to test the performance of these face recognition algorithms. However, while existing publicly-available face databases contain face images with a wide variety of covariates such as poses, illumination, gestures and face occlusions but there is no dummy face database is available in public domain. The contributions of this paper are: i) Preparation of dummy face database of 50 subjects ii) Testing of face recognition algorithms on the dummy face database, iii) Critical analysis of four algorithms on dummy face database.

Prediction of pain intensity using multimedia data

Increased medical expenditure, improper treatment, decreased productivity and above all inadequate pain, assessment is the global challenging problem, causing a substantial burden on the individual, their family, and society which often leads to loss of interest in life. The facial expression variation generally reflects clue for the occurrence of pain. This offers a vital aspect for non-verbal patients who are not in a position to rate their pain intensity level. The hypothesis proposes the developing of a multimodal machine based tool which will assess pain intensity. A framework has been designed to meet up with these specific issues which involve extracting features from the face and genes involved in pain. Patients suffering from pain require a comprehensive assessment to be conducted for proper diagnosis. The contributions of this paper are fourfold: Firstly, this paper provides an efficient approach to computational pain quantification. Secondly, it investigates the medical practitioner perception to pain along with the readings of the various tools available. Thirdly, the psychological aspect is taken into consideration to predict how pain is perceived by observers and experts (physicians). Fourthly, genes involved in pain and no pain conditions are taken and classified. Three large databases of spontaneous pain expressions are used i.e., McMaster UNBC Pain Archive database, self-prepared database and the other BioVid heat pain database of pain to verify the accuracy and robustness of the system. The methodology achieves 87% accuracy rate for classification of frames amid four levels of pain intensity. Designing intelligent computing systems with the given methodology will certainly improve the quality of life of patients.