Abstract
Background Estimation is a common computer vision task, used for segmenting moving objects in video streams. This can be useful as a pre-processing step, isolating regions of interest for more complicated algorithms performing detection, recognition, and identification tasks, in order to reduce overall computation time. This is especially important in the context of embedded systems like smart cameras, which may need to process images with constrained computational resources. This work focuses on accelerating SuperBE, a superpixel-based background estimation algorithm that was designed for simplicity and reducing computational complexity while maintaining state-of-the-art levels of accuracy. We explore both software and hardware acceleration opportunities, converting the original algorithm into a greyscale, integer-only version, and using Hardware/Software Co-design to develop hardware acceleration components on FPGA fabric that assist a software processor. We achieved a 4.4× speed improvement with the software optimisations alone, and a 2× speed improvement with the hardware optimisations alone. When combined, these led to a 9× speed improvement on a Cyclone V System-on-Chip, delivering almost 38 fps on 320 × 240 resolution images.
Highlights
Many computer vision applications rely on scanning an image by applying a sliding window across the image, whether it is a simple filter operation or a more complex object detection and recognition task
The key reason to focus on SuperBE is that it is more accurate that algorithms like Gaussian Mixture Model (GMM) without introducing the significant computation costs of more modern background estimation approaches
While SuperBE as an algorithm is more complex and slower than GMM or similar methods previously accelerated, it has an average Percentage of Wrong Classifications (PWC) of 1.75%, much better than the 4% error rate expected from GMM
Summary
Many computer vision applications rely on scanning an image by applying a sliding window across the image, whether it is a simple filter operation or a more complex object detection and recognition task. Background estimation ( known as background subtraction, background modelling, or foreground detection) is a popular method of segmenting images in order to isolate foreground regions of interest This allows for further analysis with subsequent algorithms, saving computation time by processing a smaller image and iterating over fewer window positions. Sequential frames in a video are usually required in order to compare frames to each other and classify similar parts of the images as background, and there is a general assumption that objects of interest are moving between frames This eliminates the applicability of background estimation in some offline image processing applications where the images contain no notion of time or may be entirely independent, but for most real-world applications there is some continuous monitoring where multiple frames of the same view are captured.
Sign-in/Register to view highlights & summary Talk to us
Join us for a 30 min session where you can share your feedback and ask us any queries you have
Schedule a callDisclaimer: All third-party content on this website/platform is and will remain the property of their respective owners and is provided on "as is" basis without any warranties, express or implied. Use of third-party content does not indicate any affiliation, sponsorship with or endorsement by them. Any references to third-party content is to identify the corresponding services and shall be considered fair use under The CopyrightLaw.
