Important Note: This article is part of the series in which TechReport.us discuss theory of Video Stream Matching.
1.1 – Introduction
Image processing is now a days a dynamic field. Information technology is making progress rapidly in up ward direction. Its role in our daily life is also increasing. Now people are more depending on information technology. As demand increasing its application are also changing. Every day new problems came for software engineers. One of the future requirements in next 30 years will be video analysis. Soon image processing replace with video processing. Most of the information now a days shot on camera and placed in the form of video. Text and Hyper Text are replacing by video. Now more news channels then news papers. Now more reports are on national geographic then in the Time magazines.[6]
VSM ( Video Stream Matcher ) takes video as data and analysis it for required Test image or Video. There is a source video which trained the system. There are two possibilities for test data. One is system got a single image then analysis it on system, if it is in source video or not. Another possibility is that it takes test data also a video and analysis if this video is the part of source video or not. Decision making ability system got with the help of different statistical models as KS Test, Variation, Mean, and Norm.
Video is a huge collection of frames. To minimize it a technique is applied called key frames extraction.[7, 8]
After that from each frame selected features are extracted. Ion this application four features are extracted. These features are histogram, edge, slope, wavelets.[6]
1.2 – Literature Review
VSM is actually a research work. For this report many research papers are studies. For covering the basics of image processing and understanding the implementation of image processing books are studies.
For research work taking help form these research papers and implementing their define algorithms.
Similarity Analysis of Video Sequences using an Artificial Neural Network
SUCHENDRA M.BHANDARKAR AND FENG CHEN
Department of Computer Science University of Georgia, USA
SPRINGER SCIENCE Group
Key Frames Extraction and Indexing for Multimedia Databases
Mohammad Ahmad, Ahmad Karmouch
School of Information Technology & Engineering (SITE)
University of Ottawa, Canada
Key Frame Extraction and Shot Retrieval Using Feature Line (NFL)
Li Zhao, Wei Qi, Stan Z Li, Shi Qiang Yang, H.J. Zhang
Department of Computer Science and Technology
Tsinghus University, Beijing, China
Microsoft Research China
Signature Verification Using Kolmogorov Smirnov Statistic
Harish Srinivasan, Sargur N. Srihari and Matthew J Beal
University at Buffalo, the Sate University of New York,
Buffalo, USA. Jan 9, 2005
MATLAB 7.0 is using for this research work, Project Coding and Implementation
1.3 – Key Elements of VSM
This implementation work is called VSM.VSM components are,
- It takes a video as source Data on which system trained.
- It gets Test data as image or as video.
- On input of both types of videos Key Frames extraction take place.
- Four features extract from Key Frames histogram, edge, slope, Wavelet.
- Statistical models Apply for decision making.
System has the ability to process the data automatically after getting input. Only data input for processing. User interfering is minimized in that system.
1.4 – Problem Statement
Now a days if you traveling through motor way then before entering on that root a person enter into the bus and start shouting a video. It takes image of every person. This is due to security reasons. After taking that video they evaluate that video. This process is totally manual. So one person see all people faces and trying to recognize if any criminal traveling through that bus. This is a teasing and not 100% efficient work. So a system needs which evaluate that video and gives required results.
Similarly in T.V station there are many videos place in database. These databases if going on to digitalized then maintaining, designing and searching from these system a application required which process the videos for required part of the video. VSM is the solution of these types of problems.
VSM and VSM research work help to find the better way to build a better application. This report is making a new area in image processing and helping to solve the future problems and also help us to see in the future.
1.5 – Proposed Solution of the problem
VSM works on videos. So solution is revolves around the Videos.
It takes input as video. Video is huge in size means consist on number of frames. Therefore Key Frames extraction approach applied on it. Key frames are frames those representing the video clip uniquely. One video may consist on many clips. So each clip has one frame respectively.
Whole video is represented by few frames called key frames. Then from these frames features extract to get useful information. Then system ( VSM ) help us for decision making. Only those features are taken which uniquely represent the frame / image or has the ability to classifying the data.
Then statistical models are applying on them in two phases. This approach is called “Dual Approach for Decision Making”. First phase gives us nearest values and second phase gives final results.
1.6 – Scope Of The Project
- Extract the Mean Frames from the video
- Extract features using algorithms
- Features extracted are histogram, edge, slope, wavelets
- Use statistical dual approach on it for decision making.
1.7 – Organization Of Report
This report is actually combing two themes. First one is research and finding and second one is implementation of that finding and research work.
Chapter two give briefing introduction of image processing concepts which are used in that work. Chapter 3 where image processing and preprocessing is discusses. In chapter 4 implementation of software and algorithm are discussed. In chapter 5 software working and in chapter 6 conclusion and future work related discussion included.
