Anzeige
Erschienen in:
2018 | OriginalPaper | Buchkapitel
Real-Time Key Frame Extraction in a Low Lighting Condition
verfasst von : Tao Shen, Zhan-Li Sun, Fu-Qiang Han, Ya-Min Wang
Erschienen in: Advances in Neural Networks – ISNN 2018
Aktivieren Sie unsere intelligente Suche, um passende Fachinhalte oder Patente zu finden.
Wählen Sie Textabschnitte aus um mit Künstlicher Intelligenz passenden Patente zu finden. powered by
Markieren Sie Textabschnitte, um KI-gestützt weitere passende Inhalte zu finden. powered by
Abstract
In a low lighting condition, the key frame extraction is an intractable problem due to the degraded video quality. Aimed at this issue, an effective real-time key frame extraction method is proposed to deal with the video captured in a low lighting condition. Firstly, we invert the image and turn the low-dark background into a fog-like scene. Then, we construct the atmospheric scattering model, and reconstruct the enhanced low-light image by removing the haze and inverting the image. Furthermore, a selection strategy based on hash criterion is designed to determine whether the transmission map needs to be recalculated to improve the processing speed. Moreover, an improved vibe algorithm is presented to model and extract foreground objects. Finally, according to the ratio of the foreground object to the whole image, we judge whether there is a moving object in a frame. The experimental results on some typical videos demonstrate the feasibility of the proposed method.