Smoke Detection on Video Sequences based on Spatio-Temporal Local Binary Patterns in Outdoor Spaces under Complicated Weather Conditions
Abstract
Purpose: Early smoke detection in outdoor scenes using video sequences is one of the important tasks in modern surveillance systems. Visual information as a result of real-time video shooting may include objects with dynamic behavior, noise of the hardware or transmission lines, as well as artefacts affected by weather conditions (for example, rain or snow, poor luminance in the morning or evening). Therefore, smoke-like regions segmented into video sequences should be finally verified. Results: An algorithm is proposed to process images with atmospheric artefacts like drizzle or fog, or images with poor luminance. A method has been studied which involves texture analysis based on spatio-temporal local binary patterns, local ternary patterns, and extended binary patterns to detect dense or transparent smoke with the following artefacts: salt-pepper noise up to 10 dB, additive white Gauss noise simulating atmospheric precipitates, image blurring and poor luminance, using Laplace filter. To classify smoke regions, histograms were applied as one of the simplest and fastest methods of image analysis. As a measure of histogram difference between two images, Kullback-Leibler divergence was used in order to provide a decision rule. Practical relevance: The developed method of smoke verification on video sequences using spatio-temporal local binary patterns and 3D extended local binary patterns provides 96-99 % of accuracy for dense smoke and 86-94 % of accuracy for transparent smoke, depending on the artefacts and noise.Published
2016-02-22
How to Cite
Favorskaya, M., & Pyataeva, A. (2016). Smoke Detection on Video Sequences based on Spatio-Temporal Local Binary Patterns in Outdoor Spaces under Complicated Weather Conditions. Information and Control Systems, (1), 16-25. https://doi.org/10.15217/issn1684-8853.2016.1.16
Issue
Section
Information processing and control