Real-time images are often captured and processed without any buffer delays. Since most real-time images are captured from many sources, the quality of the image resolution may vary. However, due to recent advancements in image information processing, there are various types of real-time image information processing techniques. Real-time image information processing may lead to high computational overhead and delays in the transmission of the images, and to overcome these limitations, deep neural networks (DNNs) and data mining techniques (DMs) may be an asset moving forward. DNNs are very popular due to big data support and automatic features selection, this will reduce the workload of scientists, and also CNN techniques will be used to increase accuracy as compared to machine learning methods.
DNNs, like convolution neural networks (CNN), Deep adversarial network (DAN), long short-term memory (LSTM), autoencoder, and deep belief networks have been used to provide real-time image information processing. Using deep neural networks, various hidden layers present will capture important features of an image or a frame. When the image is captured on a real-time basis, it can be processed by deep neural networks more efficiently and effectively. However, there may be significant performance pressure on the processing and evaluation of real-time high-resolution and multi-resolution images. This special issue provides an exemplary forum for researchers to discuss theories and ideas associated with real-time image information processing using deep neural networks methods and optimization algorithms. Also, this special issue discusses all the challenges and limitations of using deep neural networks models in real-time image information processing.
This special issue aims to receive high-quality papers that extend the current state of the art
with innovative ideas and solutions in the broad area of utilization of deep neural networks in
real-time image information processing. Contributions may present and solve open research
problems, integrate efficient novel solutions, present performance evaluations, and compare
new methods with existing solutions. Theoretical as well as experimental studies for typical
and newly emerging convergence technologies and use cases enabled by recent advances are
encouraged. High-quality review papers are also welcome.
Potential topics include but are not limited to the following:
- DNNs/DMs-based real-time image information processing techniques
- Intelligent learning algorithms for real-time image reconstruction and enhancement
- Real-time image security and privacy using DNNs/DMs
- Federated learning methodologies used in real-time image information processing
- Enhancement of real-time images in remote sensing applications using DNNs/DMs
- Quality of Experience (QoE) and Quality of Service (QoS) for real-time image information
- DNNs/DMs pattern recognition in real-time image information processing
- Evaluation of enhanced real-time images using DNNs/DMs methods
- Computational-based DNNs/DMs models for detection of abnormalities in real-time
- New objective functions of DNNs/DMs for real-time image reconstruction
- Performance analysis of semantic segmentation of images using DNNs/DMs algorithms
- Sports and Arts image information processing using DNNs/DMs
- Limitations of DNNs/DMs and hybrid models for real-time image information processing