Colorimetric recognition for urinalysis dipsticks based on quadratic discriminant analysis

Detection of biomarkers in urine sample is often conducted by use of dipsticks, which provides a qualitative result. Urinalysis involving image recognition and data processing has becoming one of the powerful tools in clinical diagnosis. This paper presents colorimetric recognition of urinalysis dipsticks based on quadratic discriminant analysis (QDA) in order to overcome the drawbacks, such as, limited detection area, seriously affected by the external light conditions etc. It can decrease the error of color space conversion by directly processing the data from the captured image using QDA. The correlation of the sRGB color space and the difference of covariance matrix of the acquired data were took into account in this discriminant analysis. The results of validation experiments by Matlab simulation show that it can effectively identify the similarity between the test and reference color on the dipsticks with the color recognition accuracy at 97.33%.