A novel Background Subtraction Scheme for in-Camera Acceleration in Thermal Imagery

Nikitakis, Antonis; Papaefstathiou, Ioannis; Makantasis, Konstantinos; Doulamis, Anastasios

doi:10.3850/9783981537079_0092

Cited by 6 publications

(3 citation statements)

References 12 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…This work extends our previous works in [34], [35]. Our method exploits GMMs with unknown number of components, which are dynamically estimated directly from the data.…”

Section: B Our Contributionmentioning

confidence: 62%

Data-Driven Background Subtraction Algorithm for In-Camera Acceleration in Thermal Imagery

Makantasis

Nikitakis

Doulamis

et al. 2018

IEEE Trans. Circuits Syst. Video Technol.

Self Cite

View full text Add to dashboard Cite

Abstract-Detection of moving objects in videos is a crucial step towards successful surveillance and monitoring applications. A key component for such tasks is called background subtraction and tries to extract regions of interest from the image background for further processing or action. For this reason, its accuracy and real-time performance is of great significance. Although, effective background subtraction methods have been proposed, only a few of them take into consideration the special characteristics of thermal imagery. In this work, we propose a background subtraction scheme, which models the thermal responses of each pixel as a mixture of Gaussians with unknown number of components. Following a Bayesian approach, our method automatically estimates the mixture structure, while simultaneously it avoids over/under fitting. The pixel density estimate is followed by an efficient and highly accurate updating mechanism, which permits our system to be automatically adapted to dynamically changing operation conditions. We propose a reference implementation of our method in reconfigurable hardware achieving both adequate performance and low power consumption. Adopting a High Level Synthesis design, demanding floating point arithmetic operations are mapped in reconfigurable hardware; demonstrating fastprototyping and on-field customization at the same time.

show abstract

“…This work extends our previous works in [34], [35]. Our method exploits GMMs with unknown number of components, which are dynamically estimated directly from the data.…”

Section: B Our Contributionmentioning

confidence: 62%

Data-Driven Background Subtraction Algorithm for In-Camera Acceleration in Thermal Imagery

Makantasis

Nikitakis

Doulamis

et al. 2018

IEEE Trans. Circuits Syst. Video Technol.

Self Cite

View full text Add to dashboard Cite

show abstract

“…To select a motion-detection method, a designer must consider memory requirements, precision, and speed suitable for the application. 3 Current research concerns accelerating the processing speed of these methods by implementing them in field programmable gate arrays (FPGAs), 4 graphics processing units (GPUs), 5 and application-specific integrated circuit (ASIC). 6 Comparisons between FPGAs and GPUs indicate that implementation with GPUs involves less development time, whereas implementation using FPGAs consumes less power (about an order of magnitude 7 ) and achieves a smaller size.…”

mentioning

confidence: 99%

“…There are several studies in the literature that highlight the benefits of using FPGAs for embedded vision systems. In the study by Nikitakis et al, 4 a background subtraction method was implemented for vision systems distributed in different FPGAs and processors. The results showed that FPGAs achieve lower power consumption and better fps performance.…”

mentioning

confidence: 99%

FPGA‐SoC implementation of an ICA‐based background subtraction method

Carrizosa-Corral

Vázquez-Cervantes

Montes

et al. 2018

Circuit Theory & Apps

View full text Add to dashboard Cite

SummaryThis paper presents the development and implementation of an independent component analysis (ICA)–based background subtraction method on a field programmable gate array (FPGA) system on a chip (SoC) with embedded processor. The use of the classic form of ICA for this purpose results in a complex implementation with high hardware resource usage for an embedded system. Therefore, an alternative version of FastICA was developed that adapts directly to the parallelism offered by the FPGA. In addition, the incorporation of this version of ICA into the motion‐detection method exploits the architecture of the FPGA‐SoC with embedded processor. This recent technology complements the parallelism of the FPGA with the general‐purpose computing capacity of the processors while maintaining low energy consumption and a compact size. The above features are appropriate for autonomous devices that handle mainly background subtraction, while a central device receives this information and performs the remaining tasks required for the application. The use of processors allows a faster implementation and integration of the device into other systems in a simple manner, while the parallelism of the FPGA increases the processing speed by accelerating intensive numeric calculations. The results obtained from the implementation that uses both FPGA and the embedded processor of the SoC show an increase of 3300% in the number of frames per second compared with an implementation that uses only the embedded processor.

show abstract

Fusing RGB and Thermal Imagery with Channel State Information for Abnormal Activity Detection Using Multimodal Bidirectional LSTM

Bakalos

Voulodimos

Doulamis

et al. 2021

Cyber-Physical Security for Critical Infrastructures Protection

Self Cite

View full text Add to dashboard Cite

In this paper, we present a multimodal deep model for detection of abnormal activity, based on bidirectional Long Short-Term Memory neural networks (LSTM). The proposed model exploits three different input modalities: RGB imagery, thermographic imagery and Channel State Information from Wi-Fi signal reflectance to estimate human intrusion and suspicious activity. The fused multimodal information is used as input in a Bidirectional LSTM, which has the benefit of being able to capture temporal interdependencies in both past and future time instances, a significant aspect in the discussed unusual activity detection scenario. We also present a Bayesian optimization framework that fine-tunes the Bidirectional LSTM parameters in an optimal manner. The proposed framework is evaluated on real-world data from a critical water infrastructure protection and monitoring scenario and the results indicate a superior performance compared to other unimodal and multimodal approaches and classification models.

show abstract

A novel Background Subtraction Scheme for in-Camera Acceleration in Thermal Imagery

Cited by 6 publications

References 12 publications

Data-Driven Background Subtraction Algorithm for In-Camera Acceleration in Thermal Imagery

Data-Driven Background Subtraction Algorithm for In-Camera Acceleration in Thermal Imagery

FPGA‐SoC implementation of an ICA‐based background subtraction method

Fusing RGB and Thermal Imagery with Channel State Information for Abnormal Activity Detection Using Multimodal Bidirectional LSTM

Contact Info

Product

Resources

About