Abstract
Video sensors with embedded compression offer significant energy savings in transmission but incur energy losses in the complexity of the encoder. Energy efficient video compression architectures for CMOS image sensors with focal-plane change detection are presented and analyzed. The compression architectures use pixel-level computational circuits to minimize energy usage by selectively processing only pixels which generate significant temporal intensity changes. Using the temporal intensity change detection to gate the operation of a differential DCT based encoder achieves nearly identical image quality to traditional systems (4dB decrease in PSNR) while reducing the amount of data that is processed by 67% and reducing overall power consumption reduction of 51%. These typical energy savings, resulting from the sparsity of motion activity in the visual scene, demonstrate the utility of focal-plane change triggered compression to surveillance vision systems.
Highlights
Video compression is among the most computationally intensive tasks in current imaging technology [1][2]
We present compression architectures that employ focal-plane change detection as a temporal processor, rather than spatial, to selectively encoded video data to reduce the power consumption in scenarios where static scenes dominate
The power consumption of the video encoding system can be divided into three parts - the energy consumed by the pixel array and ADC to acquire the image, the energy consumed by the digital processor to process and compress the data, and the energy required to transmit the resulting bit stream, Eframe~EsensorzNpixelsEADC zNopsEDSPzNbitsETX
Summary
Video compression is among the most computationally intensive tasks in current imaging technology [1][2]. Advanced compression schemes like H.264 provide, simultaneously, high compression rates and low visual distortion. Implementation, is costly in both terms of power consumption and hardware complexity and is ill-suited for mobile applications. For situations requiring a low power, long term wireless vision sensor, an alternative approach is justified for two reasons. First in many sensor applications, like surveillance, scenes are predominantly static necessitating a sensor platform that does not expend energy processing irrelevant data. The tradeoffs between bandwidth, power and visual quality are different. The first two must be prioritized with the provision of maintaining an image sufficient to identify events and subjects of interest
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.
