Digital color photographs blur measurement investigation

• Main Author: Chew, Chin Woo.
• Other Authors: Chan Syin
• Format: Final Year Project
• Language: English
• Published: 2009
• Subjects: DRNTU::Engineering::Computer science and engineering::Computing methodologies::Image processing and computer vision
• Online Access: http://hdl.handle.net/10356/16949
• Institution: Nanyang Technological University

With the prevalence of digital cameras and the growing popularity of photo sharing website like Facebook, it has resulted in ever-increasing number of photos stored by average consumer. This has raised the demand for the organization of the photos collection and the assessment of image quality. This experimental report provides an investigation and analysis of various blur measure methods and finds the optimum methods for the purpose of classifying photos based on the quality of the images. The blur measurement algorithm used in this project is based on finding the average spread of edge width in an image. Photos are classified according to the blur measure value calculated. The aim of this project is to investigate on various factors that will affect the performance of the algorithm and to design new methods that can assist in classification of images. From the results, it was found that standardizing the image’s dimension reduces processing time and not affecting image classification outcome. It is seen that HSV-converted image does not work well with the blur measurement algorithm. New method of recovering photos with depth of field can be done by using auto focus region detection together with the calculation of edge counts in the image. In conclusion, there is a marked improvement in the modified blur measurement algorithm compared to the original algorithm done by previous student. This blur measurement function can be implemented with the new method to get better classification of photos in term of their blur rating. It must be reminded, in reality, low complexity and efficiency in performance plays important roles in the designing of the blur measurement function and its classification.