| 江朝晖,朱家微,邰其乐,岳旭东,洪石兰.基于热成像的动态多目标温度智能监测仪设计[J].安徽建筑大学学报,2022,30(2):58-62 |
| 基于热成像的动态多目标温度智能监测仪设计 |
| Design of Dynamic Multi-Target Intelligent Scanner Based on Thermal Imaging |
| |
| DOI： |
| 中文关键词: 热红外成像 温度监测 MCU TensorFlow Lite 目标识别 |
| 英文关键词: hermal infrared imaging temperature monitoring MCU TensorFlow Lite object identification |
| 基金项目:安徽省高校自然科学研究重大项目（KJ2019ZD20）；智慧农业技术与装备安徽省重点实验室自主创新研究基金（APKLSATE 2019X002） |
|
| 摘要点击次数: 4700 |
| 全文下载次数: 4275 |
| 中文摘要: |
| 为了满足小型室内场所人员体温自动监测的需求，设计一款低成本的动态多目标温度智能监测仪。 采用高分辨率微型热成像模块 Lepton 3.5 获取热红外图像，在 STM32H743 上搭载 TensorFlow Lite 深度学习框 架，监测热红外图像中的人体并标记其温度，重点基于“滑动窗口”技术设计新的人体识别算法，解决了由 于 MCU 算力不足导致的视频卡顿问题。实验结果表明，目标识别的准确率高达 99.7%，监测多个动态目标温 度时视频流畅。该温度监测仪体积小，成本低，监测目标多，适用于家庭、办公室和实验室等场所。 |
| 英文摘要: |
| A low-cost dynamic multi-target scanner is designed to monitor body temperature automatically in small indoor places. The authors use Lepton 3. 5，a high-resolution micro thermal imaging module，to acquire thermal infrared images，and apply deep learning framework TensorFlow Lite on STM32H743 to detect the human body in thermal infrared images and mark its temperature. A new algorithm for body recognition based on the sliding window technology is designed to fix video lag caused by insufficient MCU computing power. The experimental results show that the recognition accuracy reaches as high as 99. 7%，and the video is smooth when monitoring multiple dynamic targets. With its small size，low cost and multi-target detection technology，the scanner is suitable for places including home，office and laboratory. |
|
查看全文
查看/发表评论 下载PDF阅读器 |
| 关闭 |
