针对化工、农业等行业的重金属检测需求，提出了基于图像颜色辨识的有色溶液浓度测定原理，利用顺序注射技术和图像辨识技术，设计了重金属6价铬水质在线监测系统，实现水质6价铬测定。按照国标法配置了9组不同浓度的溶液，设计了CCD相机检测池，利用6价铬水质自动在线监测仪和CCD相机在国标方法检测的同时进行图像采集。对图片数据进一步分析，建立了RGB特征值与浓度的多元线性回归模型。实验结果表明，在测量浓度为0~1.0 mg/L范围内，相对标准偏差小于2%，检出限为0.006 mg/L，基于图像辨识的检测系统具有较高的精确度。系统运行稳定，可精准、高效分析不同地表水样的6价铬，为环境保护和污水排放提供了新的技术支持。

A concentration determination principle for color solution is developed based on image recognition,in order to meet the demand for detection of heavy metals in the chemical industry and agriculture.By utilizing sequential injection and image recognition technologies,a water quality on-line detection system for chromium(Ⅵ) is designed to realize the determination of chromium(Ⅵ) in water.In accordance with China's national standard method,nine groups of solutions with different concentrations are configured,and a CCD camera detection pool is designed and established.A self-designed automatic on-line water quality detection monitor for chromium(Ⅵ) is used to detect these nine groups of solutions,while CCD camera is simultaneously used for image acquisition.Upon the analysis on the collected images,a shadow compensation method is proposed to improve the image quality,and a multiple linear regression model between RGB eigenvalues and concentrations is established.Experimental results show that the repeatability of this system is less than 1.5% in the range of 0.0-1 mg·L^-1 and the detection limit is 0.006 mg·L^-1.Compared with China's national standard method,this system has higher precision.The system runs stably,and can analyze accurately and efficiently chromium(Ⅵ) in different surface water samples,which provides new technical support for environmental protection and sewage discharge.

李文(1975-),男,博士,副教授,研究方向为机器人技术、智能传感器,lw@ncut.edu.cn