CN106991679A - One kind quantifies recognition methods based on cloud platform urine test paper physical signs - Google Patents

Abstract

The invention is a method for quantifying and identifying physiological indicators of urine test paper based on a cloud platform. The urine test paper includes three parts: a two-dimensional code, a standard color card, and a test paper. After using the test paper for urine test, take pictures with the mobile phone, automatically collect the test paper images, upload them to the cloud platform, and the cloud platform integrates image recognition methods, and returns the results to the mobile phone after processing. The recognition method first performs Harris corner detection calibration image; then converts the image space from RGB to CIELAB, calculates the color difference between the color of the test paper and the color block of the standard color card, and then uses the KNN nearest neighbor algorithm to classify the color of the test paper into the colorimetric The category corresponding to the card can automatically complete the analysis of the measurement indicators of the test paper. The method is fast and simple, avoids the problems of ambient light interference and low accuracy of human eye recognition, and is low in cost. It can also be used to screen patients with early chronic kidney disease and diabetes to reduce the risk of diseases.

Description

A quantitative identification method for physiological indicators of urine test strips based on cloud platform

technical field

The invention belongs to the category of image recognition, and in particular relates to a quantitative recognition method for physiological indicators of urine test strips based on a cloud platform, which can be used for hospital emergency treatment and screening of patients with early and chronic kidney disease and diabetes.

Background technique

In the medical field, the diagnosis of a disease usually requires the measurement of various parameters, most of which come from the test results of the samples, and the use of test strips is a widely used detection method. Among the many tests, urinalysis is an important means of detecting many diseases, and it is of great significance in clinical diagnosis. At present, many electronic devices have entered the medical field, and urine analyzers have also appeared one after another. Urine analyzers generally have a dedicated reagent strip, and the reagent strip row contains several reagent blocks, which are equidistantly distributed on the test paper. The test paper blocks are coated with chemical reagents specially used to detect corresponding items, and the number of reagent blocks determines the number of test items. The reagent cubes are coated with chemicals that measure the corresponding item, and react in color when in contact with urine. According to the degree of color change of the reagent block, the concentration of the corresponding substance in the urine can be judged and the detection result can be obtained.

Urinalysis is widely used in the diagnosis of urinary system diseases, curative effect observation and prognosis, and diagnosis of metabolic system diseases. Using traditional methods for urine testing is complicated and takes a long time. Although urine analyzers can quickly and accurately measure various indicators of urine, most urine analyzers are too expensive and have a limited detection range, so matching test strips must be used for detection. However, the traditional human eye contrast recognition method is inefficient and easily affected by the subjective factors of the staff. In order to solve these problems, a cloud-based urine test strip image recognition method is proposed, which can be used on mobile devices. Detection, to help users perform urine self-test at home, adopts the method of urine dry chemical analysis, use mobile device camera to scan test paper and color card to get semi-quantitative measurement results, with low cost, fast detection, easy to use, etc. advantage.

Contents of the invention

The present invention aims to provide a method for quantifying and identifying physiological indicators of urine test paper based on a cloud platform, which is low in cost, simple and intuitive, shortens the detection time, and has a high accuracy rate.

The present invention is to provide a method for quantifying and identifying physiological indicators of urine test paper based on a cloud platform, comprising the following steps: step 1, obtaining the image of urine test paper; step 2, detecting and calibrating images of Harris corner points; step 3, converting the image space from RGB to CIELAB, calculate the color difference between the urine test image and the standard color card; step 4, use the KNN nearest neighbor algorithm to classify the color of the test paper into the corresponding category of the color card, and automatically analyze the concentration of the indicator.

The urine test test paper includes a two-dimensional code part, a standard color card part, and a detection test paper part, wherein, by scanning the two-dimensional code, the link URL of the test paper can be opened, and the user instructions, manufacturer, standard chromatogram, examples, etc. can be viewed, serving for taking pictures Identify the color of each indicator reaction.

The urine test paper image recognition method can recognize the color and concentration of the test paper blocks of different manufacturers, different models, and different detection items.

Description of drawings

Fig. 1 is a structural schematic diagram of the urine test paper.

Fig. 2 is a flow chart of the image recognition method for urine test paper based on the cloud platform.

Figure 3 is a flowchart of Harris corner detection.

Figure 4 is a flow chart of CIELAB spatial color difference calculation.

detailed description

The cloud platform-based urine test paper image recognition method provided by the present invention will be described in detail below in conjunction with the accompanying drawings.

Fig. 1 is a structural schematic diagram of the urine test paper, which includes three parts: a two-dimensional code part (1), a standard color card part (2), and a detection test paper part (3). By scanning the QR code part (1), you can obtain the model, instruction, manufacturer, standard color spectrum, examples, etc. of the test paper; the standard color card part (2) is used to judge the color similarity between the test paper (2) and the standard color card when testing degree; the detection test paper part is used for color reaction with the urine to be tested.

When using, first drop urine on the test paper, and after the reaction is complete, first measure the color of the test paper with human eyes to determine its reference value in the standard color card.

Fig. 2 is the flowchart of the urine test paper image recognition method based on cloud platform, comprises the following steps: step 1, obtains the image (4) of urine test paper and standard color card; Step 2, Harris corner detection calibration image (5); Step 3. Transfer the image space from RGB to CIELAB (6), calculate the color difference between the urine test image and the standard color card (7); step 4, use the KNN nearest neighbor algorithm to classify the color of the test paper into the corresponding category of the color card (8), Automatic analysis of indicator concentrations.

The image (4) of obtaining the urine test paper and the standard color card refers to the image of the test paper and the standard color card after the reaction is collected by the camera of the mobile device. In order to avoid the interference of ambient light, the camera should simultaneously collect the urine test paper and the color card image, After the camera captures the original image, the original image is first compressed, and the new pixel is the average value of the original surrounding average value, which has the effect of average value filtering.

Fig. 3 is the flow chart of Harris corner detection, described Harris corner detection calibration image (5), is used to eliminate the impact of shaking in the shooting process on image acquisition result, the basic mathematical formula of Harris corner detection is: , including the following steps:

step one: Represents the moving window, according to which the first-order Gaussian partial derivatives Ix and Iy on the X-axis and Y-axis of the image are calculated;

Step 2: Get the values of Ix^2 , Iy^2 and Ix*Iy according to the results of the first step;

Step 3: In the second step of Gaussian blur, get the three values Sxx, Syy, Sxy;

Step 4: Calculate the eigenvalues of the matrix based on the Harris matrix of the pixel ;

Where the Harris matrix is:

Step 5: Calculate the R value of each pixel;

Its formula is: , ,

Step 6: Use a 3*3 or 5*5 window to achieve non-maximum suppression;

Step 7: According to the corner point detection result, find the matching point of the color block matrix in the extracted corner point, and perform adaptive sampling;

After the image is calibrated, the calibrated image needs to be uploaded to the server. After receiving the image data uploaded by the client, the server section generates a color correction model according to the color data of the color correction color card and the standard color card data in the image. Perform brightness adjustments and color corrections for accurate color reproduction. In terms of color correction, ICCS (the Image Color Correction System) is used for color restoration, which is the specific process described in Figure 4.

Fig. 4 is the CIELAB space chromatic aberration calculation flow chart, described image space is transferred to CIELAB (6) from RGB, because the original image data collected by camera is stored in RGB format, and calculating CIELAB space chromatic aberration needs to convert RGB color data To the CIELAB color space, but the RGB color space cannot be directly converted into the CIELAB color space, it needs to be converted into the CIEXYZ color space first and then converted into the CIELAB color space, namely: RGB—XYZ—LAB, including the following steps:

Step 1: Calculate the specific values of the R, G, and B pixel channels, and use the gamma function to edit the color in a non-linear tone to improve the image contrast;

The value ranges of R, G, and B pixel channels are all [0,255]. The formula for calculating the R, G, and B pixel values is as follows:

in

Step 2: Convert the RGB color space to the color space. The formula for converting the color space to the CIEXYZ color space is as follows:

where M=[0.4124,0.3576,0.1805,0.2126,0.7152,0.0722,0.0193,0.1192,0.9505]

Step 3: Convert CIEXYZ color space to CIELAB color space, the formula is as follows:

Step 4: Euclidean space distance is used to evaluate the CIELAB space color difference color similarity, corresponding to the calculation of the color difference (7) between the urine test image and the standard color card, the formula is as follows:

The KNN nearest neighbor algorithm classifies the color of the test paper into the corresponding category of the color chart (8), and classifies the color of the sampled test paper into the sample category of the color chart in the CIELAB color space, and first selects the sample of the color chart in the CIELAB color space According to the KNN nearest neighbor algorithm, the paper samples to be tested are classified into the color category corresponding to the selected nearest neighbor color, so as to complete the analysis of the test paper index data.

Claims (6)

1. a kind of urine test paper physical signs quantifies recognition methods, it is characterised in that comprise the following steps：

Step 1, acquisition urine test paper image；

Step 2, Harris Corner Detection uncalibrated images；

Step 3, image space from RGB gone into CIELAB, calculate the aberration of urine examination image and standard color card；

Step 4, using KNN nearest neighbor algorithms by test paper color classification to the corresponding classification of colorimetric card, automatic analytic index concentration.

2. a kind of urine test paper image-recognizing method based on cloud platform according to claim 1, it is characterised in that：It is described Urine test paper contain Quick Response Code part, standard color card part, Test paper part, wherein, can be beaten by scanning Quick Response Code The link network address of the test paper is opened, operation instruction, manufacturer, standard colour chart, example etc. is checked, each index reaction of identification of taking pictures is served Color afterwards.

3. a kind of urine test paper image-recognizing method based on cloud platform according to claim 1, it is characterised in that：It is described Step 3 in conversion image space image is gone to CIEXYZ spaces from rgb space first, return again to CIELAB spaces.

4. a kind of urine test paper image-recognizing method based on cloud platform according to claim 1, it is characterised in that：It is described Step 4 refer to the test paper color classification of sampling in CIELAB color spaces into colorimetric card sample class, first choose Colorimetric card sample of color in CIELAB color spaces, and by test paper sample classification to choose the corresponding color class of closest color In not, so as to complete the parsing of test paper achievement data.

5. a kind of urine test paper image-recognizing method based on cloud platform according to claim 1, it is characterised in that：It is described Urine test paper image-recognizing method can recognize different manufacturers, different model, the indicator paper block color of different detection project and its Concentration.

6. a kind of urine test paper image-recognizing method based on cloud platform according to claim 1, it is characterised in that：It is described Color of image identification based on mobile phone photograph obtain image, cloud platform complete colour recognition after return the result to mobile phone.