International Journal of Computer Applications (0975 – 888) Volume 47– No.19, June 2012 1 Overview of Scene Change Detection - Application to Watermarking Dolley Shukla Associate Professor , Dept. of IT Shri Shankaracharya College of Engg .& Tech. Manisha Sharma Professor & HoD. E & Tc Bhilai Institute of Technology, Durg ABSTRACT With the advent of digital video and digital broadcasting, watermarking of video data has been one of the important issues. Scene change detection (SCD) is one of several fundamental problems in the design of video watermarking. It is the first step towards automatic segmentation, annotation, and indexing of video data. SCD is also used in other aspects of watermarking i.e. copy-protection, copyright-protection of videos. Therefore, for the copy-protection of video using watermarking, scene change detection is an important step. In this paper we provide classification and comparison of different scene change detection techniques & algorithms. Keywords Video watermarking, abrupt change, scene change detection, compressed & uncompressed video 1. INTRODUCTION Detection of scene changes play important roles in video processing with many applications ranging from watermarking, video indexing, video summarization to object tracking and video content management. Scene change detection is an operation that divides video data into physical shots[1]. Over the last three decades, scene change detection has been widely studied and researched. As a result, many scene change detection techniques have been proposed and published in the literature. For our convenience of surveying existing research in this subject area, all these algorithms and techniques can be broadly classified as operating on decompressed data (pixel domain), or working directly on the compressed data (compressed domain)[2]. Scene changes are divided into two types: Abrupt scene change and Gradual scene change. Abrupt scene changes result from editing “cuts” and detecting them is called cut detection either by colour histogram comparison on the uncompressed video or by DCT coefficient comparison. Gradual scene changes result from chromatic edits, spatial edits and combined edits. Gradual scene changes include special effects like zoom, camera pan, dissolve and fade in/out, etc [3]. 2. FEATURES OF SCENE BOUNDARY DETECTION Almost all scene change detection algorithms reduce the large dimensionality of the video domain by extracting a small number of features from each video frame. These are extracted either from the whole frame or from a subset of it, which is called a region of interest (ROI). Such features include[4]: 2.1 Luminance/colour The simplest feature that can be used to characterize a ROI is its average gray scale luminance. This, however, is susceptible to illumination changes. A better choice is to use some statistics of the values in a colour space . 2.2 Luminance/colour histogram A richer feature for a ROI is the gray scale or colours histogram. It is quite discriminates, easy to compute and mostly insensitive to translational, rotational and zooming camera motion, for the above reasons it is widely used. 2.3 Image edges An obvious choice of feature is edge information in a ROI. Edges can be used as is, be combined into objects or used to extract ROI statistics. They are invariant to illumination changes and most motion, and they correspond somewhat to the human visual perception. Their main disadvantage is computational cost, noise sensitivity and high dimensionality. 2.4 Transform coefficients (DFT, DCT, Wavelet): These are a classic way to describe the texture of a ROI. The DCT coefficients are also present in MPEG encoded video streams or files. Their greatest problem is that they are generally not invariant to camera zoom. 2.5 Other features A number of other features are used in the literature, such as the colour angiogram. 2.6 Multiple features Many algorithms extract several types of features either to use them in combination or for subsequent processing and analysis. 3. REVIEW ON SCENE CHANGE DETECTION With the rapid growth of multimedia production and distribution, more and more people, concern the problem on copy protection of multimedia data. This problem become obvious as the explosive growth of the internet in recent few years. Video watermarking requires the scene change detection in the first step. Hence, this section presents a review of different scene change detection techniques of videos. Scene change detection algorithms are based on the pixel differences, compressed(MPEG- 2)domains, temporal segmentation luminance histograms based framework for temporal segmentation, sudden scene change detection for MPEG-2 compressed video, algorithm using direct edge information extraction from MPEG video data is used.