Region of interest from "focus region" in EXIF data

• Main Author: Arundas Thulasidas Chandraleeba.
• Other Authors: Chia Liang Tien
• Format: Final Year Project
• Language: English
• Published: 2012
• Subjects: DRNTU::Engineering::Computer science and engineering::Computing methodologies::Image processing and computer vision
• Online Access: http://hdl.handle.net/10356/48617
• Institution: Nanyang Technological University

Digital image processing is a promising field with thriving active research for developing new algorithms that improve a computer’s ability to process images and detect meaningful information in them. As computers are currently unable to look at an entire image and identify its contents the way humans can, these algorithms are deemed very useful in automating processes such as face detection, identification of regions of interest, etc. This project attempts to create a user-friendly software program named ‘FocusWhere’ which can be used to identify the regions of interest in a photograph using various methods. The program is designed in such a way that it enables a user to both easily apply the methods on a single photograph to visualize its results, as well as quickly apply them to a large number of photographs. There are three different methods made available in FocusWhere to identify the regions of interest. Before using any of the methods, a technique known as edge detection is first applied to the image as this greatly reduces the information in the images while retaining the necessary structural features. The methods are discussed in detail in the fourth chapter of this report. The software program also attempts to utilize any available EXIF data in the images to improve the accuracy of its region of interest identification methods. Certain models from Nikon® and Canon® brands are supported by the program. The EXIF data contains valuable information about the parts of an image where the camera focused when taking the photograph. This information is used to determine the optimal location in the image for applying the methods. It was found that two of the available methods give the best results for a wide range of images. The conclusion from the test results is that different methods can be utilized for different kinds of images, but no single method could be found to work for all images. This report attempts to explain all the above in a systematic order, beginning with a short introduction to FocusWhere and discussion of the user interface, followed by a detailed discussion of the various algorithms used in the software program, and concluding with the test results and comments, as well as recommendations for future work.