二维Otsu和改进区域生长法的荧光免疫层析试条浓度的定量检测

目前荧光免疫层析试条的定量检测方法绝大多数采用光电反射法，该方法需要对试条进行高精度的定位并且需要复杂的传动装置。利用CMOS图像传感器获取荧光免疫层析试条的图像信息，采用图像分割方法用于荧光免疫层析试条检测线和质控线进行自动识别，首先对采集到的原始图像进行增强预处理，并进行初步分割。然后选定处理后图像的种子点并通过改进的二维Otsu法选取生长阈值对图像进行区域生长，以实现对图像检测线和质控线的分割。结果表明该方法能够在目标和背景对比度很低的情况下，将图像检测线和质控线的轮廓清晰的分割出来，且区域一致性测度和对比度都比较理想。该方法检测不同浓度的试条，得到的特征值重复性小于5%，线性度大于0.99，准确度大于0.98。实验证明用该方法对荧光免疫层析试条进行定量检测是可行的，且较对比仪器具有更宽的检测范围。

Fluorescent Immune-chromatographic Strip Quantitative Detection Based on Two-Dimensional Otsu Method and Region Growth Algorithm

Existing immunofluorescence chromatographic quantitative detection technology methods are mostly based on photoelectric reflex method. High precision for the strip position,and complex strip transmission unit are required in this method. In this paper,image of fluorescent immune-chromatographic strip was captured by the CMOS image sensor. A method for automatic recognition of test line and control line was proposed,In this method, seed points was chosen after the image was enhanced and initial segmented. Then the growth threshold was set via the improved two-dimensional Otsu method. Finally,the region growth algorithm was implemented to segment test line and control line. Experimental results showed that the algorithm could precisely segment the test line and con?trol line,and finally realized the quantitative detection of fluorescent immune-chromatographic strip. The experi?ments proved that the segmented test line and control line is clear,complete and had excellent performance,even in conditions of a low contrast or a changeable background. The proposed method indexes on quantitative evaluation of the segmentation result,such as Uniformity Measure(UM),Regional Contrast(CR),were both well. In the quantita?tively detect step,the coefficient variation of the measured characteristic values was less than 5%and the linearity was mare than 0.99. The experimental results shown that the method was feasible for the quantitative detection of fluorescent immune-chromatographic strip,and had wider detection limit than the comparing instrument.