IoT-Based Finger-Vein Biometric Recognition using Repeated Line Tracking and CLAHE Algorithm for Vault Security System Access

Human error is something that cannot be avoided especially when dealing with security systems. Hence, there are some major developments on technology to minimize these phenomena. From changing the key to lock the door into finger prints, or from using pin numbers to open vaults into using biometric recognition such as Iris, face, veins, etc. the developments on these security systems brings such massive impacts in future technology. Thus, the vein usage that is used to open a vault lock could be considered useful due to the characteristic of hidden biometrics due to its position inside the skin and its characteristics is that it could only be visible when flashed by lights. In this research, the main discussion is about the process of getting and storing vein images using image processing and compare them with the recently taken image to open vault lock. The sub discussion is about the accuracy of the comparation image and explain of the result of the accuracy. The procedure of getting the output of the best quality on the captured image is through image processing. due to the low quality of used CMOS camera, the only way to make the output image readable and detectable through edge detection is through image processing where the original color image needs to be grayscaled first then through image selection (RoI), also through the filtering process (Bilateral) and then to CLAHE algorithm and last will be processed into edge detection (Canny). The mechanism process also needs to be focused on where there are some sensors and actuators to support the tools to work properly. The main sensor that needs to be focused on is IR sensor where it is functioned as finger detection when the finger is inserted through the finger hole. There are also actuators to support the output of the compared results as indicator whether it is identical or not. The outcome of this research turns out to be a big success where the results mostly identified as Identical with an overall accuracy of 80.33% and could open the vault lock without a problem.