The Structural Similarity Can Identify the Presence of Noise in Video Data from Unmanned Vehicles

Uloženo v:
Podrobná bibliografie
Vydáno v:Journal of Imaging vol. 11, no. 11 (2025), p. 375-398
Hlavní autor: Anzor, Orazaev
Další autoři: Lyakhov Pavel, Andreev Valery, Butusov Denis
Vydáno:
MDPI AG
Témata:
On-line přístup:Citation/Abstract
Full Text + Graphics
Full Text - PDF
Tagy: Přidat tag
Žádné tagy, Buďte první, kdo vytvoří štítek k tomuto záznamu!
Popis
Abstrakt:This paper proposes a method for detecting distorted frames in video footage recorded by an unmanned vehicle. The proposed detection method is performed by analyzing a sequence of video frames, utilizing the contrast aspect of the structural similarity index between previous and current frames. This approach allows for the detection of distortions in the video caused by various types of noise. The scientific novelty lies in the targeted adaptation of the SSIM component to the task of real interframe analysis in conditions of shooting from an unmanned vehicle, in the absence of a reference. The three videos were considered during the simulation. They were distorted by random significant impulse noise, Gaussian noise, and mixed noise. Every 100th frame of the experimental video was subjected to distortion with increasing density. An additional measure was introduced to provide a more accurate assessment of distortion detection quality. This measure is based on the average absolute difference in similarity between video frames. The developed approach allows for effective identification of distortions and is of significant importance for monitoring systems and video data analysis, particularly in footage obtained from unmanned vehicles, where video quality is critical for subsequent processing and analysis.
ISSN:2313-433X
DOI:10.3390/jimaging11110375
Zdroj:Advanced Technologies & Aerospace Database