Photo album - organizing photographs, understanding aesthetic

• Main Author: Chua, Jie Hong.
• Other Authors: Chia Liang Tien
• Format: Final Year Project
• Language: English
• Published: 2011
• Subjects: DRNTU::Engineering::Computer science and engineering::Computing methodologies::Pattern recognition
• Online Access: http://hdl.handle.net/10356/46464
• Institution: Nanyang Technological University

Data Mining is an analytic process of analyzing and exploring new patterns, from a large data set. This discovery of patterns and viewpoints of behavior which were previously unnoticed, will allow the researched data to be used in prediction and for application to businesses, or for finding out the relationships between given data sets. The objective for this project is to understand the aesthetic of photographs using basic data mining techniques on available EXIF information. Time-difference Clustering, Hierarchical Clustering, K-Means Clustering, Auto-Focus Positions Detection and Canny Edge Detection are techniques that are applied in this project to analyze and process information mined from EXIF in an image. Time-difference Clustering focuses on separating photographs into different clusters according to their difference in time, computed in seconds while Hierarchical Clustering is a technique that displays the result of cluster analysis in the form of a hierarchy. K-Means further illustrates the classification of a given set of data through a certain number of clusters by associating it to the nearest centroid (average) through a series of recursive process. Auto-Focus Positions Detection explores into the readings of data from an important tag in EXIF and then breaking up the data to retrieve the actual Auto-Focus Position of the image captured. Canny Edge Detection aims to detect edges and to suppress noise at the same time. Edges characterize object boundaries and are therefore useful for segmentation, registration, and identification of objects in a scene. Both techniques are then combined together to process whether the captured Auto-Focus position of an image is accurate, based on the subject in the image. The richness in the amount of information contained in the images and the applications adapted to process these images has reckoned this project to be an in-depth, interesting and useful research.