Date of Award


Document Type


Degree Name

Master of Science (MS)


Computer Science

First Advisor

Dr. Quiming Zhu

Second Advisor

Dr. Zhengxin Chen

Third Advisor

Dr. Abhishek Parakh


People have access to a tremendous amount of video nowadays, both on television and Internet. The amount of video that a viewer has to choose from is so large that it is infeasible for a human to go through it all to find a video of interest. Organizing video into categories will make the process of large number of videos much faster and improves the ease of access. A profile created by observing the rate at which the contents of video frame changes helps in categorization of videos in different types. The experiments we conducted on three types of videos (News, Sports, and Music) show that a profile built on a set of frame transition parameter measurements could be applied to automatically distinguish the types of these videos.

We have researched a way to automatically characterize videos into their respected video type, such as a news, music, or sports video clips, by comparing the content value transitions among the video frames. The objective of this research is to see if some measurements extracted from frame transitions are used to show the differences between different categories of videos. In other words, we want to see if such kind of values and measurements can be used to tell different kind of videos or the genre of videos, e.g., with respect to the authors. Our program extracts the statistical data from the video frames based on the histograms of the grayscale pixel intensity changes in the frame transitions. A variety of videos were tested to categorize them using the extracted signatures from these frame transition profiles. The signatures extracted presents a problem of classification that can be addressed using the machine learning algorithms. Time complexity of the evaluation is decreased when compared to other methods in video classification as the video is processed in a single step where all the features are extracted and analysis is performed on the obtained signatures. This provides a simple approach in classifying the videos, additional signatures will be extracted to create a more efficient profiling system to better reveal the nature and characteristics of the video categorization.


A Thesis Presented to the Department of Computer Science And the Faculty of the Graduate College University of Nebraska In Partial Fulfillment Of the Requirements of the Degree Master of Science University of Nebraska at Omaha. Copyright 2015 Abhiram Reddy Gaddampalli.

Files over 3MB may be slow to open. For best results, right-click and select "save as..."