计算机科学
现场可编程门阵列
JPEG格式
计算机硬件
比特流
无线
图像压缩
数据压缩
嵌入式系统
帧速率
压缩比
JPEG 2000
实时计算
图像处理
算法
人工智能
图像(数学)
解码方法
电信
工程类
汽车工程
内燃机
作者
Andreas Bay Madsen,Rasmus Faurskov,Ali Sahafi
标识
DOI:10.1109/ist59124.2023.10355693
摘要
In Wireless Capsule Endoscopy (WCE), image compression can be used to reduce the necessary bandwidth and power consumption for the wireless transmission, allowing for a longer battery life for a given image resolution and frame rate. In this paper an FPGA implementation of the JPEG-LS compression algorithm is developed for use in Wireless Capsule Endoscopy with the aim of investigating the hardware resource utilization and power consumption. The pipelined HDL implementation is designed to interface directly with a camera module at its input and a transmitting device at its output, where the compressed bitstream is stored in a buffer. The full system is intended to run on the Lattice iCE40UP5K low power FPGA and the considered camera module is the OV7670 from OmniVision. Tests performed on a gastro-intestinal image dataset, representative of the use case, showed an average compression ratio of 2.84. An estimated power consumption of 10.35 mW is reported along with a utilization of 27860 4-bit Look-Up Tables (LUT4) and 136.5 KiB block-RAM. The most resource demanding module is the output buffer for the bitstream, using more than 50 percent of the consumed LUT4. The Look-Up Table consumption exceeds that of the intended target FPGA and further work is needed to optimize the design.
科研通智能强力驱动
Strongly Powered by AbleSci AI