Optimal object detection and tracking in occluded video using DNN and gravitational search algorithm
- PDF / 6,203,875 Bytes
- 20 Pages / 595.276 x 790.866 pts Page_size
- 106 Downloads / 238 Views
(0123456789().,-volV)(0123456789(). ,- volV)
FOUNDATIONS
Optimal object detection and tracking in occluded video using DNN and gravitational search algorithm T. Mahalingam1 • M. Subramoniam1
Springer-Verlag GmbH Germany, part of Springer Nature 2020
Abstract Moving object tracking is an effective optimization procedure based on the impermanent relevant information associated with the original frames. Suggesting a method with efficient accuracy in convoluted atmospheres is a difficulty for scientists in the area of research study. In this research, powerful object detection and movement tracking videos are proposed. Here, we are considering the input video sequence PETS and Hall monitor videos. Initially, the background and foreground separations are done by modified kernel fuzzy c-means algorithm. The object detection and tracking are done by gravitational search algorithm-based deep belief neural network. The implementation will be in MATLAB. The effectiveness of the recommended strategy is assessed with means of precision, recall, F-measure, FPR, FNR, PWC, FAR, similarity, specificity, and accuracy. From the experimental results, the proposed work outperforms the state of artwork. Here, the proposed method attains maximum precision and recall value for both PETS and Hall monitor video when compared to the existing algorithm. Keywords Kernel fuzzy c-means algorithm Background and foreground separation Deep belief neural network Gravitational search algorithm
1 Introduction Nowadays, the surveillance systems are more popular and necessary to interpret the info about the prospect frameworks in reconnaissance video such as an observing period, the number of identified objects, and a performance of objects (Sabirin and Munchurl 2012). In video processing, three steps are included such as detection, classification, and tracking. The pictorial object discovery and tracking are the two imperative factors of video analysis in multicam observation (Zhang et al. 2015a; Cai et al. 2010). In visual tracking, some of the problems occur in video processing such as non-rigid shapes, expression variances, occlusions, illumination uncertainties, cluttered scenes, and low resolution (Chen et al. 2015). These are the difficult Communicated by A. Di Nola. & T. Mahalingam [email protected] M. Subramoniam [email protected] 1
activities in video refinement and computer vision during recognition and tracking of progressing objects from a video scene. Detection and tracking for motion objects, detection in a video is the progression of finding dissimilar object locations which are going to background (Subudhi et al. 2011). A most generally made use of technique as detection of motion objects is background reduction (delBlanco et al. 2012) and also optical flow technique. The optical flow technique presents noticeable improvements of a motion object among the frames that determine the accelerations and positions of each point of a video frame. It is better for non-rigid object evaluation. Utilizing optical flow ev
Data Loading...
