Felix Ohms defended his Bachelor Thesis on Integrated Quality Assessment of mobile touchless Fingerprint Capturing using Android Smartphones

 

In recent years fingerprint technology has become a widespread means of securing data, objects and locations. This is largely due to the ease of using it, the low chance of losing the authenticating key, a finger, compared to a key or password, and the hardships that come with mounting an attack on such a system.

There is however still room to improve certain aspects, like the usability. Previously, acquiring fingerprints required specialised sensors which mandated direct contact with the fingertips. This made the acquisition fairly cumbersome, as the user needed to touch a specific surface in a specific way for the sensor to produce usable data. This is even more of an issue when fingerprints should be acquired from more than one finger, as they could either be done in sequence, which would dramatically increase the needed time, or require a larger sensor with a shape permitting the comfortable placement of all fingers for varying hand sizes.

Due to these factors attention has recently been given to touchless acquisition of fingerprints, which allows capturing many fingerprints at once with only a simple camera. Touchless acquisition presents its own unique set of challenges as matching algorithms need highlighted minutiae, well defined areas and high contrasts. Therefore any image captured by a normal camera needs a significant amount of preprocessing to meet these criteria before it can be processed a matching algorithm.
A smartphone application aimed at solving exactly this has been developed in a previous Master Thesis.

This project expands upon that application by designing and integrating a logging solution with a modern approach to logging design and development. Logging guidelines and practices are developed alongside the solution. These aid in keeping the logging statements and quality consistent across the application.

The logging solution is then used to gather data about the applications behaviour, specifically the processing pipeline.
This data is analysed and used to point to weak spots in the application regarding execution time and accuracy.
The main two weak points regarding the pipelines performance are then further analysed to reveal their causes as inefficient utilisation of morphological operations due to the chosen structuring element.
An additional weak point of the performance is introduced when images of a right hand are processed, as these take significantly longer in one of the required processing steps.

Changing the morphological operations to use a simpler structuring element decreases the processing time of the pipeline by 10s, which correlates to a 90% reduction in used processing time. Analysis of the improved pipeline shows no significant negative impact on the quality of the resulting images.