Non-Invasive Diabetes Mellitus Detection Using Facial Block Color
S. Sathyavathi1, K. R. Baskaran2, S. Kavitha3

1S. Sathyavathi, Department of Information Technology, Kumaraguru College of Technology, Coimbatore (Tamil Nadu), India.
2Dr. K. R. Baskaran, Department of Computer Science and Engineering, Kumaraguru College of Technology, Coimbatore (Tamil Nadu), India.
3S. Kavitha, Department of Information Technology, Kumaraguru College of Technology, Coimbatore (Tamil Nadu), India.
Manuscript received on 13 December 2018 | Revised Manuscript received on 24 December 2018 | Manuscript Published on 09 January 2019 | PP: 304-306 | Volume-7 Issue-4S November 2018 | Retrieval Number: E1982017519/19©BEIESP
Open Access | Editorial and Publishing Policies | Cite | Mendeley | Indexing and Abstracting
© The Authors. Blue Eyes Intelligence Engineering and Sciences Publication (BEIESP). This is an open access article under the CC-BY-NC-ND license (

Abstract: Diabetes Mellitus (DM) is a condition in which glucose level in the body is much higher than the normal. The traditional way to diagnosis DM is Fasting Plasma Glucose (FPG) test. As this method is slightly painful and uncomfortable several another method which are more comfortable and non-invasive are found. In this paper, we propose a new non-invasive method to detect DM based on facial block color features using various classification algorithms. Facial images are first captured using a specially designed non-invasive device, and calibrated to ensure consistency in feature extraction and analysis. Four facial blocks are extracted automatically from face image and used to represent a face features. A facial color gamut is constructed with six color centroids (red, yellow, light yellow, gloss, deep red, and black) to compute a facial color feature vector, characterizing each facial block. Finally, the features are classified using J48. ForJ48, two sub dictionaries, a Healthy facial color features sub dictionary and DM facial color features sub dictionary, are employed in the classification process. Apart from this we also use ZeroR, Support vector machine (SVM)[8] ,J48 to determine the accuracy, precision and recall using the data set that comprises of healthy and DM samples. Finally, we compare all these algorithms and choose the efficient one using its accuracy level.
Keywords: Non-Invasive, Algorithm Efficiency, Health Enhancement.
Scope of the Article: Algorithm Engineering