IFE – Mechanical Engineering

IFE – Biomedical Engineering

IFE – Pattern Recognition

For_students
Object detection in scenes recorded by a mobile device

Author: Karol Matusiak BSc

Supervisor: Piotr Skulimowski PhD

Distinction in the competition for the best thesis engineering done in the academic year 2010/2011, organized by the Polish Association of Electric Engineers (SEP) and the Faculty of Electrical, Electronic, Computer and Control Engineering of the Technical University of Lodz.

The aim of bachelor’s thesis was to design and implement an application which enables identification of urban objects in the images recorded by the built-in mobile phone camera. The images identification methods demanding relatively low computing complexity were analyzed and the requirements of designed algorithm were determined. On the basis of the analysis it was decided to use the Scale-Invariant Feature Transform. The algorithm provides the detection of stable characteristic features of images and builds descriptors which are to a considerable degree independent from the conditions of image recording, such as: rotation, noise, scale and brightness changes. As the target mobile platform a group of developed mobile phones with operating system Symbian was selected. The program was written with usage of modified by author Nokia Computer Vision Library, which contains a number of functions useful in image processing and analysis. The program structure is based on a standard application template with graphical user interface. The recording of image is done through a digital camera built in the mobile device. The program enables to build a database with patterns and to compare recorded images with patterns, using a modified nearest neighbor classification. The application tests were conducted on 8 urban objects, whose patterns were created and stored in the database. Then 24 test photos were compared against the database contents. The test results are considered to be satisfactory in the case of 67% of test images, whereas in the remaining cases the reasons of misinterpretation are indicated. In the summary, the application performance assessment and the possibilities of its improvement and development are proposed. The algorithm finds application in the range of fields, including city or tourist guide, the systems of virtual reality or as an aid to navigation in the urbanized area.