文章基本信息

标题：Object Detection Using Background Subtraction Tolerating Sudden Background Variations
本地全文：下载
作者：Aruna A S ; Deepthi K
期刊名称：International Journal of Computer Science and Information Technologies
电子版ISSN：0975-9646
出版年度：2017
卷号：8
期号：2
页码：146-148
出版社：TechScience Publications
摘要：Background subtraction is a technique in the fieldof image processing where in an image’s foreground isextracted for further processing. Low –rank and sparserepresentation method, which make few specific assumptionabout the background have really attracted wide attention inbackground modelling. Background subtraction is a widelyused approach for detecting moving object in videos. The aimof this approach is to detect the moving objects from thedifference between the current frame and reference frame. Itconstitutes of two terms, a low-rank matrix, which is thebackground and a structured sparse outlier matrix known asthe foreground. Generally, an image region of interest areobjects in its foreground. Both the background andforeground frames are converted into grey scale. After thestage of image pre-processing, object localization is required.In order to do that comparison of the two frames are done.The pixels below a predefined threshold value are convertedto white where the new object is present. The white area iscropped and compared with the already trained patterns toidentify the objects by pattern matching. Since the applicationof this technique is in the forest areas the object identificationmainly involves the recognition of animals both indigenous orvisitors in a habitat. The main advantage is that it is able totolerate sudden background variations like change in weatherconditions or turn on/off lights.
关键词：background substitution; pattern recognition;training object detection