CN-116559082-B - Sensing device, detection system, and urine detection system

CN116559082BCN 116559082 BCN116559082 BCN 116559082BCN-116559082-B

Abstract

The device comprises a filter assembly, a detection assembly and a urine detection system, wherein the filter assembly is used for coding incident light to obtain imaging information of each sensing channel, the detection assembly is used for detecting the imaging information and generating a rectangular gray image array, the rectangular gray image array is used for being input into a detection model to obtain a detection result by using the output result of the detection model, the detection model has a mapping relation between the rectangular gray image array and the detection result, and the detection result comprises components of an object to be detected or components of the object to be detected and contents of all the components. The sensing device of the embodiment of the disclosure directly detects to obtain the rectangular gray image array to input the detection model to obtain the detection result, has the advantages of low cost, small volume, simple structure and high precision, and improves the measurement precision.

Inventors

BAO JIE
HUAI BINGXIN
Request for anonymity
LIU XIAOHU

Assignees

清华大学
芯视界（北京）科技有限公司

Dates

Publication Date: 20260505
Application Date: 20230516

Claims (15)

1. A sensing device, the device comprising: The filter assembly is used for encoding the incident light to obtain imaging information of each sensing channel, the imaging information comprises light intensity values of the incident light, the filter assembly comprises a plurality of different types of filter plates, each filter plate corresponds to one sensing channel, and different filter plates can encode the incident light to obtain different imaging information; A detection component for detecting the imaging information and generating a rectangular gray image array, wherein the rectangular gray image array is used for being input into a detection model to obtain a detection result by using the output result of the detection model, the detection model has a mapping relation between the rectangular gray image array and the detection result, the detection result comprises components of an object to be detected or the components of the object to be detected and the content of each component, The rectangular gray image array comprises an array form of T rows and P columns formed by a plurality of rectangular areas, and T, P is an integer larger than 0.
2. The apparatus of claim 1, wherein the detection model is configured to determine the detection result according to the rectangular gray scale image array and the mapping relationship.
3. The apparatus of claim 1, wherein the detection model is further configured to pre-process the rectangular gray scale image array, wherein the pre-processing means comprises at least one of: averaging corresponding pixels of a plurality of the rectangular gray image arrays; And correcting the light intensity non-uniformity of each pixel of the rectangular gray image array.
4. The apparatus of claim 1, wherein the detection model is derived based on at least one of least squares, neural networks, support vector machines, naive bayes classification, decision trees, k-nearest neighbor algorithms, linear discriminant analysis, linear regression, logistic regression, classification and regression trees, learning vector quantization, bagging methods, and random forests.
5. The apparatus of claim 1, wherein if the detection model is implemented based on a least squares method, the detection model is configured to: averaging the light intensities of the corresponding rectangular areas of the rectangular gray image arrays to obtain a plurality of average light intensity values; Splicing the plurality of average light intensity values to obtain an intensity vector; And carrying out least square operation by taking the intensity vector as input, and taking an operation result as the detection result, wherein the detection result comprises the content of each component.
6. The apparatus of claim 1, wherein if the detection model is implemented based on a neural network, the detection model is configured to: Extracting image features of the rectangular gray image array; And performing convolution operation and full connection operation on the extracted image features for multiple times, and outputting a detection result.
7. The apparatus of any one of claims 1-6, wherein the types of filters include at least one of a super surface filter type, a photonic crystal filter type, a perovskite quantum dot filter type, and a colloidal quantum dot filter type, each filter type including a plurality of different types, The detection component comprises at least one element of a complementary metal oxide semiconductor element, a charge coupled element, an ultraviolet detection element and an InGaAs near infrared detection element.
8. The device of any one of claims 1 to 6, wherein in the filter assembly, the filter is a colloidal quantum dot filter, each colloidal quantum dot filter has a different spectral transmission relationship, the filter encodes incident light together based on the spectral transmission relationship and a spectral sensitivity relationship of a detection assembly corresponding to each filter, so as to obtain imaging information of the incident light, and the spectral sensitivity relationship represents a relationship between light responsivity and light wavelength.
9. The apparatus according to any one of claims 1 to 6, wherein the rectangular gray scale image array includes a plurality of rectangular areas, each filter corresponds to one rectangular area, and each rectangular area includes a plurality of pixels.
10. The apparatus of any one of claims 1-6, wherein the filter assembly is determined by: selecting a plurality of filters with different numbers from N filters to form a plurality of filter components, wherein each filter in the N filters has different spectral transmission relations, the N filters can encode incident light in a target wavelength range, and N is a positive integer; Selecting the minimum number of filter plates from detection results corresponding to filter plate assemblies with different numbers of filter plates from detection results corresponding to the first preset detection results, wherein the minimum number is used as the number of filter plates of the filter plate assemblies, the filter plate assemblies are used for encoding incident light into imaging information, the imaging information comprises light intensity values of the incident light, the filter plate assemblies comprise a plurality of different types of filter plates, and the different filter plates can encode the incident light to obtain imaging information; determining the combination mode of the minimum number of continuous distribution or jump distribution from N filter plates for multiple times; And selecting a combination mode of filter plate combinations corresponding to the optimal detection result in the second preset detection result from detection results corresponding to different filter plate combinations with the same number of filter plates as the combination mode of the filter plates in the filter plate assembly, wherein each filter plate combination comprises the minimum number of filter plates, and the types and/or arrangement modes of the filter plates in each filter plate combination are different.
11. A detection system, the detection system comprising: the sensing device according to any one of claims 1 to 10; The light source is used for emitting detection light; A reaction component for interacting with the test object to produce a color change; the detection light emitted by the light source irradiates the reaction component to obtain one or more of transmitted light, reflected light or fluorescence, and the irradiated light is incident to the filter component; The data processing assembly is used for obtaining a detection result according to the rectangular gray image array generated by the sensing device by using a detection model, wherein the rectangular gray image array is used for being input into the detection model so as to obtain the detection result by using the output result of the detection model, the detection model has a mapping relation between the rectangular gray image array and the detection result, and the detection result comprises components of an object to be detected or the components of the object to be detected and the content of each component.
12. The system of claim 11, wherein the data processing component is further configured to: Acquiring a first rectangular gray image array and a second rectangular gray image array output by the sensing device, wherein the first rectangular gray image array is a rectangular gray image array output by the sensing device when the reaction component is not added with an object to be detected, and the second rectangular gray image array is a rectangular gray image array output by the sensing device when the reaction component is added with the object to be detected; subtracting the intensities of the corresponding pixels of the second rectangular gray image array and the first rectangular gray image array to obtain a third rectangular gray image array; and inputting the third rectangular gray image array into a detection model, and obtaining a detection result of components of the object to be detected or the components of the object to be detected and the content of each component by using the output result of the detection model, wherein the detection model has a mapping relation between the rectangular gray image array and the detection result.
13. A urine detection system comprising the sensing device of any one of claims 1 to 10 or the detection system of any one of claims 11 to 12.
14. The urine detection system according to claim 13, wherein the urine detection system is configured to detect at least one of glucose content, nitrite content, urobilinogen, uroketone body, urobilirubin, uroprotein, urinary erythrocytes, leukocytes, and epithelial cells in urine to be detected, The reaction component of the urine detection system is a reflective component.
15. The urine detection system of claim 13, wherein the filter assembly of the sensing device is capable of encoding 450nm to 670nm of incident light, the number of filters in the filter assembly being 20, and the detection assembly is made of complementary metal oxide semiconductors.

Description

Sensing device, detection system, and urine detection system Technical Field The present disclosure relates to the field of detection technology, and in particular, to a sensing device, a detection system, and a urine detection system. Background The spectrum sensing method can determine the composition and content of the object to be detected according to the color change caused by physical and chemical reactions, and has wide application in the fields of medicine, environmental monitoring, agriculture and the like. The existing spectrum sensing scheme for color detection has the problems of large volume, low precision, complex structure, high cost and the like. Disclosure of Invention According to an aspect of the present disclosure, there is provided a sensing device, the device comprising: The filter assembly is used for encoding the incident light to obtain imaging information of each sensing channel, the imaging information comprises light intensity values of the incident light, the filter assembly comprises a plurality of different types of filter plates, each filter plate corresponds to one sensing channel, and different filter plates can encode the incident light to obtain different imaging information; The detection module is used for detecting the imaging information and generating a rectangular gray image array, the rectangular gray image array is used for being input into a detection model to obtain a detection result by using the output result of the detection model, the detection model has a mapping relation between the rectangular gray image array and the detection result, and the detection result comprises components of an object to be detected or components of the object to be detected and contents of all the components. In a possible implementation manner, the detection model is used for determining the detection result according to the rectangular gray scale image array and the mapping relation. In a possible implementation manner, the detection model is further used for preprocessing the rectangular gray scale image array, wherein the preprocessing mode comprises at least one of the following steps: averaging corresponding pixels of a plurality of the rectangular gray image arrays; And correcting the light intensity non-uniformity of each pixel of the rectangular gray image array. In one possible implementation, the detection model is obtained based on at least one of least squares, neural networks, support vector machines, naive bayes classification, decision trees, k-nearest neighbor algorithms, linear discriminant analysis, linear regression, logistic regression, classification and regression trees, learning vector quantization, bagging methods, and random forests. In one possible implementation, if the detection model is implemented based on a least squares method, the detection model is used to: averaging the light intensities of the corresponding rectangular areas of the rectangular gray image arrays to obtain a plurality of average light intensity values; Splicing the plurality of average light intensity values to obtain an intensity vector; And carrying out least square operation by taking the intensity vector as input, and taking an operation result as the detection result, wherein the detection result comprises the content of each component. In one possible implementation, if the detection model is implemented based on a neural network, the detection model is used to: Extracting image features of the rectangular gray image array; And performing convolution operation and full connection operation on the extracted image features for multiple times, and outputting a detection result. In one possible embodiment, the types of filters include at least one of a super surface filter type, a photonic crystal filter type, a perovskite quantum dot filter type, a colloidal quantum dot filter type, each filter type including a plurality of different types, The detection component comprises at least one element of a complementary metal oxide semiconductor element, a charge coupled element, an ultraviolet detection element and an InGaAs near infrared detection element. In a possible implementation manner, in the filter combination, the filters are colloid quantum dot filters, each colloid quantum dot filter has a different spectral transmission relationship, the filters encode incident light together based on the spectral transmission relationship and a spectral sensitivity relationship of a detection component corresponding to each filter, so as to obtain imaging information of the incident light, and the spectral sensitivity relationship represents a relationship between light responsivity and light wavelength. In one possible implementation, the rectangular gray image array includes a plurality of rectangular regions, each of the filter segments corresponds to one of the rectangular regions, and each of the rectangular regions includes a plurality of pixels. In one possible embodiment, th