【作者】 花一然；

【导师】 张校亮； 李晓春； 周明；

【作者基本信息】 太原理工大学 ， 生物医学工程（专业学位）， 2022， 硕士

【摘要】 数字图片比色分析技术通过使用RGB、HSV、CMYK等色度空间对显色产物的色彩信息量化,相较传统目视比色法减少了大量主观误差。而智能手机兼备数字图片比色分析技术的图像传感器、图像处理设备和图像分析软件等条件,并在2022年世界普及率高达59%,因此通过调用摄像头采集图片结合开源Android Studio开发软件实现生化分子定性、定量检测成为检测领域前沿方向。但在实际研发中发现,数字图片比色分析法容易受到外界环境光的光照强度变化和光照色彩偏移影响,许多科研团队通过设计出独特的外部装置隔绝环境光,并在内部补恒定光进行校正,但容易出现补光光斑、装置不便携、对不同手机机型兼容性差等问题。也有部分科研团队仅使用智能手机应用程序检测,但存在检测指标单一,操作步骤繁琐等情况。因此,本文通过研究环境光校正比色分析理论相关文献,提出一种基于智能手机和Android应用程序的数字图片比色分析系统,在不借助任何外部装置条件下,通过使用环境光校正算法结合数字图片阵列式比色分析技术对环境光的光照强度变化和色彩偏移以及智能手机自身产生的阴影校正,并实现了对九种尿液标志物的一键式同时定量检测。本文主要研究内容如下:1.设计了一种阵列式试纸条布局,一方面从实际应用角度考虑,对比单列试纸条更方便取样;另一方面是增加了用于校正的标准黑色和白色,目的在于配合探究改进后的校正算法进行逐行色度值校正,解决因手机拍照时自身产生逐行加深的阴影带来的影响。2.通过多次实验改进环境光校正算法,首先提取检测框中心区域的RGB平均值,意在解决边缘色差并减小中心区域显色不均的影响;为证明校正算法的可行性与准确性,设计了不同色系在不同条件下的验证实验进行论证;同时将校正后的RGB值转换成HSV值,用于匹配色调变化明显的显色反应体系进行多参数定量检测。3.实现了智能手机的Android操作系统下开发环境的搭建与整体功能的开发,将本研究环境光校正阵列式比色分析方法开发成独立、可兼容不同智能手机的Android应用程序,并在华为、VIVO、OPPO智能手机上调试测试。4.为验证环境光校正阵列式比色分析系统的准确性与实用性,实现了对九项尿液指标标准品的定量检测,其中通过对亚硝酸盐、肌酐和葡萄糖的检测结果与紫外分光光度计对比后一致性分别达到0.9995、0.9950和0.9921。另外分别将环境光校正阵列式比色分析系统对微量白蛋白、肌酐的检测结果与商用产品进行对比,相较于商用产品的定性半定量阈值判定,该系统具有更高的准确性;最后对华为、VIVO智能手机进行结果差异性对比,证明开发的应用程序有良好兼容性以支持不同用户使用,说明本研究有潜在广阔的商用前景。更多还原

【Abstract】 Smartphones,equipped with image sensors,image processing equipment,and image analysis software for digital image colorimetric analysis technology,have achieved a penetration rate as high as 59% globally in 2022.The digital picture colorimetric analysis technology quantifies the colour information of colour rendering products by using RGB,HSV,CMYK and other chromaticity spaces,which reduces subjective errors compared with the traditional visual colorimetric method.Therefore,calling the camera to capture images combined with open-source Android Studio development software to achieve qualitative and quantitative detection of biochemical molecules has become a cutting-edge direction in object detection.However,in the actual research and development,it was found that the digital picture colorimetric analysis method is easily affected by the light intensity change of external ambient light and light colour shift.Many research teams have designed unique external devices to isolate ambient light and create constant light internally for correction.However,they are prone to patchy light spots,non-portable devices,and poor compatibility with different cell phone models.Some research teams use only smartphone applications for testing,but there are single indicators for testing and cumbersome operation steps.Therefore,we propose a digital picture colorimetric analysis system based on smartphones and Android applications by extensively studying the literature related to the theory of colorimetric analysis with ambient light correction.We use an ambient light correction algorithm combined with digital picture array colorimetric analysis technology to collaborate for light intensity variations and colour shifts of ambient light and shadows generated by the smartphone itself without using external devices.In that case,we achieve the simultaneous quantitative detection of nine urine markers in one click.The primary contributions are as follows:1.We designed an array test strip layout.On the one hand,it is more convenient to take samples than single-column test strips from the practical application point of view.On the other hand,the standard black and white used for calibration were added,aiming to cooperate with exploring the improved calibration algorithm for row-by-row chromaticity value correction to solve the impact caused by the row-by-row deepening shadows generated by the cell phone itself when taking pictures.2.We improved the ambient light correction algorithm through several experiments.Firstly,we extracted the average RGB values in the center of the detection frame,which is intended to solve the edge colour difference and reduce the influence of uneven colour rendering in the central region.In order to prove the feasibility and accuracy of the calibration algorithm,we designed validation experiments for different colour systems under different conditions;Simultaneously,the corrected RGB values were converted into HSV values for matching colour rendering reaction systems with significant changes for multi-parameter quantitative detection.3.We realized the construction of the development environment and the development of the overall functions under the Android operating system for smartphones.We developed this study’s ambient light-corrected array-based colorimetric analysis method into an independent Android application compatible with different smartphones.Finally,we debugged and tested on Huawei,VIVO and OPPO smartphones.4.To verify the accuracy and practicality of the ambient light-corrected array colorimetric analysis system,we realized the quantitative detection of nine urine marker standards,in which the consistency of the detection results of nitrite and creatinine reached0.9995 and 0.9950 after comparing with UV spectrophotometer,respectively.In addition,we compared the detection results of the ambient light-corrected array colorimetric analysis system for microalbumin and creatinine with those of commercial products,and our system has higher accuracy compared with the qualitative and semi-quantitative threshold determination of commercial products.Finally,we compared the results of Huawei and VIVO smartphones to demonstrate the compatibility of the developed application to support different users and validate that this study has potential broad commercial prospects.更多还原