Video Summarization Based on Gaussian Mixture Model and Kernel Support Vector Machine for Forest Fire Detection
B. Pushpa1, M. Kamarasan2
1B. Pushpa, Department of Computer and Information Science Annamalai University.
2M. Kamarasan, Department of computer and Information Science Annamalai University.
Manuscript received on September 22, 2019. | Revised Manuscript received on October 20, 2019. | Manuscript published on October 30, 2019. | PP: 1827-1833 | Volume-9 Issue-1, October 2019 | Retrieval Number: A1442109119/2019©BEIESP | DOI: 10.35940/ijeat.A1442.109119
Open Access | Ethics and Policies | Cite | Mendeley
© 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: Exponential growth in the generation of multimedia data especially videos resulted to the development of video summarization concept. The summary of the videos offers a collection of frames which precisely define the video content in a considerably compacted form. Video summarization models find its applicability in various domains especially surveillance. This paper intends to develop a video summarization technique for the application of forest fire detection. The proposed method involves a set of processes namely convert frames, key frame extraction, feature extraction and classification. Here, a Merged Gaussian Mixture Model (MGMM) is applied for the process of extracting key frames and kernel support vector machine (KSVM) is employed for classifying a frame into normal frame and forest fire frame. The simulation analysis is performed on the forest fire video files from FIRESENSE database and the results are assessed under several dimensions. The final outcome proves the efficiency of the presented MGMM-KSVM model in a considerable way.
Keywords: KSVM Classification; Forest fire; Keyframe extraction; MGMM.