A fusion of dolphin swarm optimization and improved sine cosine algorithm for automatic detection and classification of objects from surveillance videos

Abstract

Surveillance is the utmost extensively used technology in the current scenario. In real-time, it is immensely applicable in all domains to monitor, identifying the moving objects, and tracking through computer vision. The object detection and classification is an important process in surveillance video. During this task, the visual appearance will change according to the viewing angles, lightening, and distance from the camera. It is necessary to improve the efficiency in real-time detection and classification of objects from surveillance videos. To obtain this, we proposed a fusion of Dolphin Swarm Optimization (DSO) and improved Sine Cosine algorithm (ISCA) based Support Vector Machine (SVM) classifier which includes the following steps: Frame differencing for foreground segmentation, Histogram of Oriented Gradients (HOG) for feature extraction, and DSO-ISCA-SVM classifier for classification. Initially, the surveillance videos are collected and the acquisition of images from the surveillance video camera. Secondly, the moving objects are detected by frame differencing in which the difference between two frames are estimated and compared with the threshold value. Then the shadow and noise are removed. Thirdly, the HOG capture local shapes through gradients. Finally, the proposed DSO-ISCA-SVM classifier accurately classifies the objects from the surveillance video, the DSO-ISCA is used to find the SVM parameters. This proposed technique effectively detects and classify the objects from the surveillance videos. The proposed technique results are compared with other existing methods. The experimental results prove that the proposed method efficiency is better than the existing methods in terms of different evaluation metrics.