ISSN: 2278 – 909X International Journal of Advanced Research in Electronics and Communication Engineering (IJARECE) Volume 1, Issue 4, October 2012 73 All Rights Reserved © 2012 IJARECE Abstract— Segmentation of moving objects in video sequences is important for multimedia aspects. Moving object segmentation is the extraction of foreground (moving object) from background. Moving object segmentation includes different steps as object detection and motion detection. The object detection and motion detection are done using different methods. This survey deals with three approaches of segmentation as region-based, boundary-based and combination of region and boundary-based methods. A comparative study is done for these methods using strengths, weakness and computational complexity. Index Terms— background subtraction, boundary based segmentation, motion detection, object detection, region based segmentation. I. INTRODUCTION Segmentation is the key concept in the image processing field especially in the image analysis process. Segmentation is used to simplify the representation of an image or video into a more relevant or informative meaningful partitions or segments or to decompose a scene into its components. The input of segmentation is a raw data including an image or a video sequence. The output is a much simpler one in which the homogenous parts are partitioned into much more simpler parts or segments. Segmentation extracts relevant information about the structure of objects from a given image or video sequence and discerns various attributes of interest from the data. These measurements or features are used for the qualitative analysis process. Segmentation is a goal dependent and subjective set-up. Segmentation of the moving objects in a video sequence is the most important and basic technology used in many real time applications. Segmentation can be useful in areas of video surveillance, security (path detection, target tracking, dynamic scene analysis), object recognition, medical field, navigation system and communication. Different visual features as color, texture, and motion, are used for achieving segmentation. Segmentation techniques are subdivided into different approaches [7] as:  Amplitude thresholding or window slicing  Component labeling segmentation  Boundary based segmentation Manuscript received Oct , 2012. Merin Antony A, PG/Scholar, Department of Computer Science and Engineering, Karunya University, Coimbatore, India. Mrs. J Anitha, Assistant Professor, Department of Computer Science and Engineering, Karunya University, Coimbatore, India.  Region based segmentation  Template matching  Texture segmentation When an object is characterized by the amplitude features, amplitude thresholding is useful. The thresholding technique can be used in various methods as background subtraction, frame difference, etc. Component labeling is an efficient method in case of segmentation of binary images. This includes pixel labeling and connectivity analysis. Boundary based segmentation and region based segmentation are most common techniques used for the segmentation process. Boundary based approach deals with the discontinuities in the images. Region based approach partition images into connected regions by grouping neighboring pixels of similar intensity. Region growing and region merging are used in this segmentation technique. In template matching, the segmentation is done by matching an image against templates from a given list. This technique is mainly used to segment busy images as, journal pages where the text detection is done using the template matching. When the objects to be segmented have textured background, texture segmentation can be used. The segmentation of objects in a video sequence is a difficult task to accomplish. Here the motion information is important. The slow moving object segmentation process is much more difficult than the segmentation of fast moving video objects. This is due to the semantic gap between the low-level visual cues as color, edge, texture, etc and high-level human interpretation of video semantics. Moving object segmentation includes the detection, tracking and extraction of the objects in motion. Detection or correspondence is keeping an account of the object that is in motion, about its course, properties, etc. Extraction is the meaningful segmentation of the moving objects from the scene. Video segmentation is used to identify the regions in a frame of video that are homogeneous with respect to any given parameter. Different features and homogeneity criteria leads to different segmentation of the same data. For example, segmentation with respect to color, texture, motion etc. Steps in moving object detection The steps for the detection of moving objects include Object Detection and Motion Detection or Motion Correspondence.  Object Detection A Survey of Moving Object Segmentation Methods Merin Antony A, J. Anitha