Enhanced Image Detection in Stereoscopic Video Using Combined Visual Saliency Techniques
DOI:
https://doi.org/10.7492/cpne3c62Keywords:
Classification, Deep learning, LSTM, Machine learning, MiDaS, Saliency, Stereoscopic, VGG, VideoAbstract
Image detection from video stereoscopes plays a critical role in applications such as object recognition, surveillance, and augmented reality. This article presents
an approach that a related method uses to extract unresolved images from a stereoscopic video sequence. Techniques for recognizing durability mimic the visual
attention of humans and allow for the identification of key regions within the framework. With the integrated spatial and temporal saliency models, the method
effectively, separates important image components, while simultaneously reducing the complexity of compensation. The results show that the accuracy in detecting
related image segments made in stereoscopic videos has been improved compared to traditional characteristic-based approaches. Experimental results show that
the combination of saliency methods surpasses traditional property-based and deep learning approaches in relation to accuracy and processing efficiency, making
it a promising solution for real-time applications. This study explores a new approach combining several methods to improve the accuracy of image detection of
stereoscopic video sequences.
