Foreground Object Detection Based on Chrominance and Texture Features with Enhancement by Canny Filter

Abstract

The foreground object detection became very important in a computer vision system and has a many applications such as recognition, object tracking, counting, classifying, home surveillance, traffic monitoring, video monitoring, medical image and in other multimedia applications. So that each of these applications needs a method for object detection, therefore, requires improving new methods and algorithms for processing this information. This paper proposes foreground objects detection approach based on the chrominance and texture features with canny enhance filter. The input is background image and current image and the output are the detecting foreground objects. The proposed approach consists of three steps: first the features extracting which are chrominance and texture features (these features are robust against to illumination changes, noise, and shadows) from a current and background image. Then, the similarity matching is computed for each feature. Finally, canny filter are used to enhance the results. Furthermore, we evaluate our approach using evaluation measures which are precision, recall, and F-measure, to give 0.922 as an average accuracy of the proposed method and with average consumption time about 0.5778923 seconds. This concludes that proposed method very efficient against the limitation of challenges and obstacles.