针对舌诊仪依赖人工、光照不均、色彩失真及缺乏动态信息等问题，提出一种基于机器视觉的自助操作暗箱舌诊仪。该仪器利用机器视觉技术和图像处理算法实现了舌象的自动定位、自动采集。高分辨率摄像技术实现3 s动态视频与高质量静态图像采集，性能测试结果显示，基于双相机几何投影标定算法，舌部识别平均时间小于0.15 s，漫反射光源提升灰度均匀度至76.74%，色彩还原度显示色差最大值为19优于行业标准值25。所开发的舌诊仪将促进中医舌诊的客观化、高效化与精准化采集。

Aiming at the problems of tongue diagnosis instruments relying on manual operation, uneven illumination, color distortion and lack of dynamic information, an automated operation black box tongue diagnosis instrument based on machine vision is proposed. This instrument utilizes machine vision technology and image processing algorithms to achieve automatic positioning and automatic acquisition of tongue manifestations. High-resolution camera technology enables the acquisition of 3-second dynamic videos and high-quality static images. The performance test results show that, based on the dual-camera geometric projection calibration algorithm, the average time for tongue recognition is less than 0.15 seconds. The diffuse reflection light source improves the gray uniformity to 76.74%, and the maximum color difference in color reproduction degree is 19, which is better than the industry standard value of 25. The developed tongue diagnosis instrument will promote the objective, efficient and precise acquisition of tongue diagnosis in traditional Chinese medicine.

中国中医科学院科技创新工程重点协同攻关项目(CI2023C009YG):基于多模态智能感知和机器学习的中医舌诊仪开发及特异性应用研究；负责人：赵宇平。国家中医药多学科交叉创新团队项目(ZYYCXTD-D-202205):中医智慧诊疗系统开发(子项目)；负责人：赵宇平。