Color recognition system based on high-sensitivity color sensitive sensor

Abstract

The invention discloses a color recognition system based on a high-sensitivity color sensitive sensor, relates to the technical field of color recognition, and solves the technical problem that in the prior art, only the connection mode of a circuit is explained, and colors in different environments cannot be finely recognized, so that a recognition result possibly has certain contingency. According to the method, actual index data of one color under different illumination intensities are collected, and the actual index data and the corresponding illumination intensities are integrated into an identification feature sequence; inputting the recognition feature sequence into a color recognition model to obtain a color label and a matched color; acquiring identification index data of the matched color, and calculating a mean value of the identification index data to obtain an identification color; putting the identification color under the same illumination intensity, obtaining feedback index data, calculating an identification deviation coefficient, and when the identification deviation coefficient is greater than a coefficient threshold value, updating the color identification model; the contingency of a single group of data is avoided, and the recognition accuracy is improved.

Description

Technical Field

[0001] The invention belongs to the field of color recognition, in particular to a color recognition system based on a high-sensitivity color-sensitive sensor. Background Art

[0002] A high-sensitivity color sensor is a sensor that can detect and identify colors with high precision and high sensitivity. A high-sensitivity color sensor can respond to a wider spectral range and can adapt to more application scenarios. In addition, a high-sensitivity color sensor has a stronger anti-interference ability, so it can accurately detect and identify colors even in complex environments. Color recognition has a wide range of applications in various fields and can be used for product quality inspection to ensure that the color of the product meets the requirements. Through automated detection and identification, production efficiency and quality can be improved. In addition, color recognition technology can be combined with other intelligent technologies to realize automation, intelligence, real-time monitoring and early warning of the production process.

[0003] The prior art (the patent application document with publication number CN107796516A) discloses a color recognition system, including a color recognition circuit, a main control circuit, a communication circuit, a power supply circuit and a voice playback circuit; the color recognition circuit, the main control circuit and the communication circuit are connected in sequence, the power supply circuit is connected to the main control circuit, and the voice playback circuit is connected to the main control circuit; the color recognition circuit includes a TCS230 color sensor, the main control circuit includes an AT89C2051 single-chip microcomputer, first to third resistors, first to third capacitors, a crystal oscillator and a reset button; the problem of complex circuits and large recognition errors is solved; however, the prior art only describes the connection method of the circuit, and cannot perform precise recognition of colors under different environments, resulting in a certain degree of randomness in the recognition results.

[0004] The present invention provides a color recognition system based on a high-sensitivity color-sensitive sensor to solve the above technical problems. Summary of the invention

[0005] The present invention aims to solve at least one of the technical problems existing in the prior art; to this end, the present invention proposes a color recognition system based on a high-sensitivity color-sensitive sensor, which is used to solve the technical problem that the prior art only describes the connection method of the circuit and cannot accurately recognize the colors in different environments, resulting in a certain degree of randomness in the recognition results.

[0006] To achieve the above-mentioned object, the first aspect of the present invention provides a color recognition system based on a high-sensitivity color-sensitive sensor, comprising: a data acquisition module, an analysis and recognition module, and a recognition feedback module; the analysis and recognition module is connected to the data acquisition module and the recognition feedback module respectively;

[0007] The data acquisition module collects light intensity through the photoelectric sensor connected thereto and transmits the light intensity to the analysis and recognition module; and collects actual index data of the color through the color-sensitive sensor connected thereto and transmits the index data to the analysis and recognition module; wherein the actual index data includes the light, hue, saturation and brightness of the color;

[0008] The analysis and recognition module determines whether the actual index data of the color can be obtained based on the light intensity; if yes, the actual index data of the color is recognized; if no, the actual index data of the corresponding color is obtained after adjusting the light intensity; and,

[0009] Based on a number of light intensities and actual indicator data, a number of recognition feature sequences are integrated; a color recognition model is called, and a number of recognition feature sequences are input into the color recognition model to obtain a number of corresponding color labels; a number of corresponding colors are matched based on a number of color labels, and a recognized color is obtained according to the recognition indicator data of a number of colors; wherein the color recognition model is constructed based on an artificial intelligence model; the recognized color is the color under a set light intensity; the recognition indicator data includes light intensity and the light, hue, saturation and brightness of the corresponding recognized color; a recognition feedback module is used to put the recognized color under a number of identical light intensities and analyze a number of feedback indicator data of the recognized color; a recognition deviation coefficient is calculated based on a number of feedback indicator data and the corresponding actual indicator data; and the color recognition model is updated based on the recognition deviation coefficient.

[0010] Preferably, judging whether actual index data of color can be obtained based on light intensity includes:

[0011] The light intensity is retrieved to determine whether the light intensity is greater than the light threshold; if so, it is determined that the actual indicator data of the color can be obtained; if not, it is determined that the actual indicator data of the color cannot be obtained.

[0012] The present invention first analyzes the existing light intensity to determine whether the actual indicator data of the color can be collected. When the actual indicator data of the color cannot be collected, the light intensity is adjusted to ensure that the collected actual indicator data can be used, thereby avoiding inaccurate collected actual indicator data due to too low light intensity.

[0013] Preferably, the plurality of identification feature sequences are integrated based on a plurality of light intensity and actual indicator data, including:

[0014] Retrieve actual indicator data of the color, and combine the actual indicator data into an indicator feature sequence;

[0015] According to the indicator feature sequence, the corresponding light intensity is matched, and the indicator feature sequence and the corresponding light intensity are combined into an identification feature sequence.

[0016] The present invention combines actual indicator data of colors into an indicator feature sequence, matches corresponding light intensity according to the indicator feature sequence, and combines the indicator feature sequence and the corresponding light intensity into an identification feature sequence; data of the same color under different light intensities can be integrated into a feature sequence, which helps to avoid the randomness of a group of data.

[0017] Preferably, obtaining the identified color according to the identification index data of a plurality of colors includes:

[0018] Retrieve a number of colors i that match color labels, and extract identification index data of color i; wherein i=1, 2, ..., n, n is a positive integer; color i corresponds to light intensity one by one;

[0019] The mean of each recognition index data is calculated by a formula, and the mean of each recognition index data is integrated into color index data; the corresponding recognition color is matched according to the color index data.

[0020] The present invention analyzes the recognition index data of several colors and recognizes the color by calculating the mean of the several recognition index data, and can combine and analyze several groups of recognition index data, which is beneficial to improving the accuracy of the recognition result.

[0021] Preferably, the color recognition model is constructed based on an artificial intelligence model, including:

[0022] Acquire standard training data, wherein the standard training data includes standard input data consistent with the content attributes of the recognition feature sequence and standard output data consistent with the content attributes of the color label;

[0023] Train the artificial intelligence model and mark the trained artificial intelligence model as a color recognition model; wherein the artificial intelligence model includes a convolutional neural network model or a BP neural network model.

[0024] Preferably, the calculation of the identification deviation coefficient based on a number of feedback indicator data and corresponding actual indicator data includes:

[0025] Retrieve the feedback light FGi, feedback hue FSi, feedback saturation FBi and feedback brightness FLi in the feedback index data; extract the actual light SGi, actual hue SSi, actual saturation SBi and actual brightness SLi in the actual index data;

[0026] The recognition deviation coefficient SPXi of the recognized color corresponding to the light intensity is calculated by the formula SPXi=α×|FGi-SGi|+β×|FSi-SSi|+γ×ln[(|FBi-SBi|)+(|FLi-SLi|)+1]; wherein α, β and γ are weight coefficients greater than 0; and ln is a logarithmic function with e as the base.

[0027] The present invention compares the feedback index data with the actual index data, calculates the recognition deviation coefficient of the recognized color corresponding to the light intensity, can provide feedback on the recognition result, is conducive to evaluating the recognition result of the color recognition model, and provides a basis for subsequent updating of the color recognition model.

[0028] Preferably, updating the color recognition model based on the recognition deviation coefficient includes:

[0029] Retrieve the recognition deviation coefficient and analyze the light intensity corresponding to the recognition deviation coefficient;

[0030] Determine whether the recognition deviation coefficient is greater than the coefficient threshold; if yes, use the feedback indicator data to update the actual indicator data in the recognition feature sequence corresponding to the light intensity; if no, continue to monitor the recognition deviation coefficient.

[0031] The present invention compares the feedback index data with the actual index data, and calculates the recognition deviation coefficient of the light intensity corresponding to the recognized color through a formula; when the recognition deviation coefficient is greater than the coefficient threshold, the color recognition model is updated, which is beneficial to improving the accuracy and reliability of the color recognition model in recognizing the color.

[0032] Preferably, the analysis and identification module communicates and / or is electrically connected to the data acquisition module and the identification feedback module respectively; the data acquisition module communicates and / or is electrically connected to the photoelectric sensor and the color-sensitive sensor respectively.

[0033] Compared with the prior art, the present invention has the following beneficial effects:

[0034] 1. The present invention collects actual indicator data of colors under different light intensities, and splices the actual indicator data into an indicator feature sequence; matches the corresponding light intensity according to the indicator feature sequence, and splices the indicator feature sequence and the corresponding light intensity into an identification feature sequence; can match the light intensity with the real-time indicator data, which is beneficial for subsequent color identification; calls a color recognition model, inputs the identification feature sequence into the color recognition model, and obtains a number of color labels; matches colors based on a number of color labels, obtains identification feature data of a number of colors, calculates the mean of each identification feature data, and analyzes the identified color based on the mean; can avoid the contingency that occurs in a single acquisition, which is beneficial for making the identified color closer to the actual color.

[0035] 2. The present invention puts the identified color under several conditions of the same light intensity to obtain several feedback index data, matches the feedback index data with the actual index data according to the light intensity, and calculates the identification deviation coefficient through a formula; when the identification deviation coefficient is greater than the coefficient threshold, the feedback index data is used to update the actual index data in the identification feature sequence corresponding to the light intensity, and the color recognition model is updated; the index data of the identified color can be fed back and the color recognition model can be updated, which is beneficial to improving the accuracy and reliability of the color recognition model in identifying the color. BRIEF DESCRIPTION OF THE DRAWINGS

[0036] In order to more clearly illustrate the embodiments of the present invention or the technical solutions in the prior art, the drawings required for use in the embodiments or the description of the prior art will be briefly introduced below. Obviously, the drawings described below are only some embodiments of the present invention. For ordinary technicians in this field, other drawings can be obtained based on these drawings without paying creative work.

[0037] Figure 1 It is a schematic diagram of the overall framework of the present invention;

[0038] Figure 2 It is a schematic diagram of the specific process steps of the present invention;

[0039] Figure 3 A schematic diagram of specific steps for color recognition of the present invention;

[0040] Figure 4 A schematic diagram of specific steps of identifying feedback in the present invention. DETAILED DESCRIPTION

[0041] The technical solution of the present invention will be clearly and completely described below in conjunction with the embodiments. Obviously, the described embodiments are only part of the embodiments of the present invention, rather than all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without creative work are within the scope of protection of the present invention.

[0042] See also Figure 1-Figure 2 , the first aspect of the present invention provides a color recognition system based on a high-sensitivity color-sensitive sensor, comprising: a data acquisition module, an analysis and recognition module, and a recognition feedback module; the analysis and recognition module is respectively connected to the data acquisition module and the recognition feedback module;

[0043] The data acquisition module collects light intensity through the photoelectric sensor connected thereto and transmits the light intensity to the analysis and recognition module; and collects actual index data of the color through the color-sensitive sensor connected thereto and transmits the index data to the analysis and recognition module; wherein the actual index data includes the light, hue, saturation and brightness of the color;

[0044] The analysis and recognition module determines whether the actual index data of the color can be obtained based on the light intensity; if yes, the actual index data of the color is recognized; if no, the actual index data of the corresponding color is obtained after adjusting the light intensity; and,

[0045] Integrate a number of identification feature sequences based on a number of light intensities and actual indicator data; call a color recognition model, input a number of identification feature sequences into the color recognition model, and obtain a number of corresponding color labels; match a number of corresponding colors based on a number of color labels, and obtain an identification color according to the identification indicator data of a number of colors; wherein the color recognition model is constructed based on an artificial intelligence model; the identification color is the color under a set light intensity; the identification indicator data includes light intensity and the light, hue, saturation and brightness of the corresponding identification color;

[0046] The recognition feedback module puts the recognized color under several conditions of the same light intensity, analyzes several feedback index data of the recognized color; calculates the recognition deviation coefficient based on the several feedback index data and the corresponding actual index data; and updates the color recognition model based on the recognition deviation coefficient.

[0047] See also Figure 3 , adjust the light intensity to obtain several actual indicator data of the corresponding color; retrieve the actual indicator data of the color, and piece the actual indicator data into an indicator feature sequence; match the corresponding light intensity according to the indicator feature sequence, and piece the indicator feature sequence and the corresponding light intensity into an identification feature sequence; call the color recognition model, input several identification feature sequences into the color recognition model, obtain several corresponding color labels, and match several colors; retrieve the color i matched with several color labels, and extract the identification indicator data of color i; wherein, i=1, 2,…, n, n is a positive integer; color i corresponds to light intensity one by one; calculate the mean of each identification indicator data through a formula, and integrate the mean of each identification indicator data into color indicator data; match the corresponding identification color according to the color indicator data.

[0048] For example: there is a color, and the actual indicator data of this color under different light intensities is retrieved; each group of actual indicator data is matched with the light intensity to identify and obtain the corresponding color label; two groups of color labels are obtained by analysis, and the corresponding identification indicator data of the two groups of colors are matched; the mean of the identification indicator data is calculated to obtain the color indicator data, and the color is identified according to the color indicator data.

[0049] See also Figure 4, put the identified color under several conditions of the same light intensity, analyze several feedback index data of the identified color; retrieve the feedback light FGi, feedback hue FSi, feedback saturation FBi and feedback brightness FLi in the feedback index data; calculate the identification deviation coefficient SPXi of the light intensity corresponding to the identified color through the formula SPXi=α×|FGi-SGi|+β×|FSi-SSi|+γ×ln[(|FBi-SBi|)+(|FLi-SLi|)+1]; wherein α, β and γ are weight coefficients greater than 0; ln is a logarithmic function with e as the base; retrieve the identification deviation coefficient, analyze the light intensity corresponding to the identification deviation coefficient; judge whether the identification deviation coefficient is greater than the coefficient threshold; if yes, use the feedback index data to update the actual index data in the identification feature sequence corresponding to the light intensity; if not, continuously monitor the identification deviation coefficient.

[0050] For example, there are feedback index data and actual index data of a color under two different light intensities, α = 0.5, β = 0.4, γ = 0.2; the coefficient threshold is 8, and the two sets of data are as follows:

[0051] The first set of data: feedback light FG1 = 645nm, feedback hue FS1 = 47, feedback saturation FB1 = 30%, feedback brightness FL1 = 76%; actual light SG1 = 650nm, actual hue FS1 = 53, actual saturation FB1 = 32%, actual brightness FL1 = 75%; the calculated recognition deviation coefficient SPX1 = 5.178, which is less than the coefficient threshold, does not need to update the color recognition model;

[0052] The second set of data: feedback light FG2=683nm, feedback hue FS2=56, feedback saturation FB2=44%, feedback brightness FL2=67%; actual light SG2=672nm, actual hue FS2=44, actual saturation FB2=58%, actual brightness FL2=78%; the recognition deviation coefficient SPX2=10.952 is calculated, which is greater than the coefficient threshold. The recognition deviation coefficient is placed in the recognition feature sequence of light intensity 2 to update the color recognition model.

[0053] Part of the data in the above formula is calculated by removing the dimension and taking its numerical value. The formula is a formula closest to the actual situation obtained by software simulation of a large amount of collected data; the preset parameters and preset thresholds in the formula are set by technical personnel in this field according to actual conditions or obtained through simulation of a large amount of data.

[0054] The working principle of the present invention is as follows: the present invention collects light intensity through a photoelectric sensor and collects actual index data of color through a color-sensitive sensor; obtains a number of actual index data corresponding to the color by adjusting the light intensity, splices the actual index data of the color into an index feature sequence, and splices the index feature sequence and the corresponding light intensity into an identification feature sequence; calls a color recognition model, inputs a number of identification feature sequences into the color recognition model, and obtains a number of color labels; matches the corresponding colors based on a number of color labels, analyzes a number of identification index data of the color, and determines the identification color by calculating the mean of each identification index data.

[0055] The present invention puts the identified color under several conditions of the same illumination intensity, analyzes several feedback index data of the identified color; calculates the identification deviation coefficient of the illumination intensity corresponding to the identified color according to the feedback light, feedback hue, feedback saturation and feedback brightness in the feedback index data; retrieves the identification deviation coefficient, and analyzes the illumination intensity corresponding to the identification deviation coefficient; when the identification deviation coefficient is greater than the coefficient threshold, the feedback index data is used to update the actual index data in the identification feature sequence corresponding to the illumination intensity, and the color recognition model is updated.

[0056] The above embodiments are only used to illustrate the technical method of the present invention rather than to limit it. Although the present invention has been described in detail with reference to the preferred embodiments, those skilled in the art should understand that the technical method of the present invention may be modified or replaced by equivalents without departing from the spirit and scope of the technical method of the present invention.

Claims

1. A color recognition system based on a high-sensitivity color sensor, include: Data acquisition module, analysis and identification module and identification feedback module; the analysis and identification module is connected to the data acquisition module and the identification feedback module respectively; it is characterized in that: The data acquisition module collects light intensity through the photoelectric sensor connected thereto and transmits the light intensity to the analysis and recognition module; and collects actual index data of the color through the color-sensitive sensor connected thereto and transmits the index data to the analysis and recognition module; wherein the actual index data includes the light, hue, saturation and brightness of the color; The analysis and recognition module determines whether the actual index data of the color can be obtained based on the light intensity; if yes, the actual index data of the color is recognized; if no, the actual index data of the corresponding color is obtained after adjusting the light intensity; and, Integrate a number of identification feature sequences based on a number of light intensities and actual indicator data; call a color recognition model, input a number of identification feature sequences into the color recognition model, and obtain a number of corresponding color labels; match a number of corresponding colors based on a number of color labels, and obtain an identification color according to the identification indicator data of a number of colors; wherein the color recognition model is constructed based on an artificial intelligence model; the identification color is the color under a set light intensity; the identification indicator data includes light intensity and the light, hue, saturation and brightness of the corresponding identification color; The recognition feedback module puts the recognized color under several conditions of the same light intensity, analyzes several feedback index data of the recognized color; calculates the recognition deviation coefficient based on the several feedback index data and the corresponding actual index data; and updates the color recognition model based on the recognition deviation coefficient.

2. A color recognition system based on a high-sensitivity color sensor according to claim 1, It is characterized in that The determining whether actual indicator data of color can be obtained based on light intensity includes: The light intensity is retrieved to determine whether the light intensity is greater than the light threshold; if so, it is determined that the actual indicator data of the color can be obtained; if not, it is determined that the actual indicator data of the color cannot be obtained.

3. A color recognition system based on a high-sensitivity color sensor according to claim 1, It is characterized in that The plurality of identification feature sequences are integrated based on a plurality of light intensity and actual index data, including: Retrieve actual indicator data of the color, and combine the actual indicator data into an indicator feature sequence; According to the indicator feature sequence, the corresponding light intensity is matched, and the indicator feature sequence and the corresponding light intensity are combined into an identification feature sequence.

4. A color recognition system based on a high-sensitivity color sensor according to claim 3, It is characterized in that The step of obtaining the identified color according to the identification index data of the plurality of colors includes: Retrieve a number of colors i that match color labels, and extract identification index data of color i; wherein i=1, 2, ..., n, n is a positive integer; color i corresponds to light intensity one by one; The mean of each recognition index data is calculated by a formula, and the mean of each recognition index data is integrated into color index data; the corresponding recognition color is matched according to the color index data.

5. A color recognition system based on a high-sensitivity color sensor according to claim 4, It is characterized in that The color recognition model is constructed based on an artificial intelligence model, including: Acquire standard training data, wherein the standard training data includes standard input data consistent with the content attributes of the recognition feature sequence and standard output data consistent with the content attributes of the color label; Train the artificial intelligence model and mark the trained artificial intelligence model as a color recognition model; wherein the artificial intelligence model includes a convolutional neural network model or a BP neural network model.

6. A color recognition system based on a high-sensitivity color sensor according to claim 1, It is characterized in that The calculation of the recognition deviation coefficient based on the plurality of feedback indicator data and the corresponding actual indicator data includes: Retrieve the feedback light FGi, feedback hue FSi, feedback saturation FBi and feedback brightness FLi in the feedback index data; extract the actual light SGi, actual hue SSi, actual saturation SBi and actual brightness SLi in the actual index data; The recognition deviation coefficient SPXi of the recognized color corresponding to the light intensity is calculated by the formula SPXi=α×|FGi-SGi|+β×|FSi-SSi|+γ×ln[(|FBi-SBi|)+(|FLi-SLi|)+1]; wherein α, β and γ are weight coefficients greater than 0; and ln is a logarithmic function with e as the base.

7. A color recognition system based on a high-sensitivity color sensor according to claim 6, It is characterized in that The updating of the color recognition model based on the recognition deviation coefficient comprises: Retrieve the recognition deviation coefficient and analyze the light intensity corresponding to the recognition deviation coefficient; Determine whether the recognition deviation coefficient is greater than the coefficient threshold; if yes, use the feedback indicator data to update the actual indicator data in the recognition feature sequence corresponding to the light intensity; if no, continue to monitor the recognition deviation coefficient.

8. A color recognition system based on a high-sensitivity color sensor according to claim 1, It is characterized in that The analysis and identification module communicates with and / or is electrically connected to the data acquisition module and the identification feedback module respectively; the data acquisition module communicates with and / or is electrically connected to the photoelectric sensor and the color-sensitive sensor respectively.

Images