Underwater Moving Object Detection by Temporal Information
Pankaj kumar, Lecturer of Information Technology, Govt. Polytechnic College, Ludhiana, Punjab.
Manuscript received on 3 August 2019. | Revised Manuscript received on 7 August 2019. | Manuscript published on 30 September 2019. | PP: 959-964 | Volume-8 Issue-3 September 2019 | Retrieval Number: C4112098319/19©BEIESP | DOI: 10.35940/ijrte.C4112.098319
Open Access | Ethics and Policies | Cite | Mendeley | Indexing and Abstracting
© The Authors. Blue Eyes Intelligence Engineering and Sciences Publication (BEIESP). This is an open access article under the CC-BY-NC-ND license (http://creativecommons.org/licenses/by-nc-nd/4.0/)
Abstract: Underwater moving object detection is a critical task for many computer vision application such as object recognizing, locating and tracing. The low accuracy rate and absence of prior knowledge learning limits its application in various underwater application. This work proposes underwater moving object detection technique based on a temporal difference technique that extends basic frame difference method to multiple frames. The proposed technique does not require any prior knowledge such as background modeling nor interaction by user such as empirical thresholds tuning. Based on continuous symmetric difference of adjacent frames, we generate full resolution saliency map of current frame to highlight moving objects with higher saliency values. This process also aids in inhibiting saliency of background also. Individual frames are obtained from the video. Frame difference is calculated of two consecutive frames. Range filters are used to get edges of object and Morphological operations are used to suppress the noise present in the foreground. The proposed algorithm is tested for performance evaluation by performing various experiments under different conditions. The testing of proposed algorithm is done by visual and statistical parameters evaluated by simulation of different videos. Versatile Experiments have done to check performance of algorithm i.e. performance of proposed algorithm in low lighting conditions, performance of proposed method in case of shadow elimination, performance of proposed method in turbulent conditions, performance of proposed method in presence of multiple objects and performance of proposed method in case of false detections. In addition, comparison with most commonly techniques for object detection like GMM and Optical Flow is also done. The proposed technique provides effective results as contrast to GMM and Optical Flow.
Keywords: Underwater Moving Object Detection, GMM, optical Flow.
Scope of the Article: Community Information Systems