当前位置: X-MOL 学术IEEE Trans. Nucl. Sci. › 论文详情
Our official English website, www.x-mol.net, welcomes your feedback! (Note: you will need to create a separate account there.)
Raw Data Processing Using Modern Hardware for Inspection of Objects in X-Ray Baggage Inspection Systems
IEEE Transactions on Nuclear Science ( IF 1.9 ) Pub Date : 2021-04-28 , DOI: 10.1109/tns.2021.3075256
S. Malarvizhi , R. Kayalvizhi , Amit Kumar , Anita Topkar

We present a novel approach for real-time implementation of raw image data processing using modern hardware for material classification and pseudocolor presentation of the images in X-ray baggage inspection systems. This implementation was done using a heterogeneous hardware comprising field-programmable gate array (FPGA) and an ARM processor available in a Xilinx FPGA board. With implementation of various algorithms in the processor system and with parallel computing done in programmable logic, an improvement in the processing speed has been achieved. For denoising of images, a Wiener filter was implemented with a high-level synthesis tool. The hardware implementation of various algorithms was verified by processing about 750 images in a Xilinx FPGA board and comparing these images with software-processed images. The processing time of a 512×640512 \times 640 pixel image was 400 ms using logic clocked at 100 MHz. Experimental results demonstrate that it is preferable to implement the Wiener filter in IP core processing rather than in the ARM as the execution is about 16 times faster in this case. The static and dynamic power consumption of the hardware was estimated to be about 2% and 92% of the total power consumption of about 1.7 W. The approach of hardware-based raw data processing could be further adopted for integration with data acquisition electronics for making the inspection systems more compact and portable with cost reduction.

中文翻译:


使用现代硬件处理 X 射线行李检查系统中的物体的原始数据处理



我们提出了一种新颖的方法,使用现代硬件实时实现原始图像数据处理,以在 X 射线行李检查系统中进行材料分类和图像的伪彩色呈现。该实现是使用异构硬件完成的,该硬件包括现场可编程门阵列 (FPGA) 和 Xilinx FPGA 板中提供的 ARM 处理器。通过在处理器系统中实现各种算法以及在可编程逻辑中完成并行计算,实现了处理速度的提高。为了对图像进行去噪,使用高级合成工具实现了维纳滤波器。通过在 Xilinx FPGA 板上处理大约 750 个图像并将这些图像与软件处理的图像进行比较,验证了各种算法的硬件实现。使用时钟频率为 100 MHz 的逻辑,处理 512×640512 × 640 像素图像的时间为 400 毫秒。实验结果表明,最好在 IP 核处理中而不是在 ARM 中实现维纳滤波器,因为在这种情况下执行速度大约快 16 倍。硬件的静态和动态功耗估计分别占总功耗约1.7 W的约2%和92%。可以进一步采用基于硬件的原始数据处理方法与数据采集电子设备集成,以实现检测系统更加紧凑、便携,同时降低了成本。
更新日期:2021-04-28
down
wechat
bug