A method and device for online automatic detection and control of the color of brown rice with germ

Abstract

The application discloses a kind of embryo rice colority on-line automatic detection and control method and device, it is related to rice processing technical field, to the problem of low efficiency and poor accuracy in prior art by observing the colority of rice sample by human eye, the present application mainly solves the evaluation and detection problem of rice colority, the specific detection target of rice is carried out image acquisition by preset illumination, and according to the data processing and calculation of image acquisition, the colority evaluation data of rice is obtained, so that the data and the degree of depth of rice are determined.The given determination data and the degree of rice germ retention and peeling have close correlation and evaluability.Compared with the colority of traditional human eye observation rice sample, the efficiency of the present application is increased by more than 2 times, and the accuracy is increased by more than 1 times.

Description

TECHNICAL FIELD

[0001] The present application relates to the technical field of rice processing, in particular to a kind of germ rice chroma identification system. BACKGROUND

[0002] Germ rice is favored by consumers in recent years as a kind of nutritious rice. The nutritional components of germ rice are mainly concentrated in the yellow germ and the bran layer on the surface of rice. The color depth (from yellow to white) of the surface layer of germ rice, i.e. chroma, is related to its nutritional components, i.e. the deeper the chroma, the higher the nutritional components, but the taste is poor; while the lighter the chroma, the lower the nutritional components, but the taste is good. Therefore, it is necessary to produce germ rice with different chroma to meet the different needs of consumers such as the elderly and children for the taste and nutrition of germ rice. In the past, the production of germ rice was manually adjusted according to the color depth of the rice sample by experience and visual observation, which was low in efficiency and poor in accuracy. With the development of automatic control technology and the progress of artificial intelligence, big data and algorithms, it is possible and necessary to conduct real-time, rapid and accurate online detection and control of the chroma of germ rice in the production process of germ rice. SUMMARY

[0003] The purpose of the present application is to solve the problem of low efficiency and poor accuracy in the prior art by observing the chroma of the rice sample by the human eye, and to propose a germ rice chroma identification system.

[0004] The technical solution adopted by the present application to solve the above technical problem is:

[0005] A kind of germ rice chroma online automatic detection and control method, comprising the following steps:

[0006] Step one: set the RGB three-primary color brightness value of the three-primary color LED light source to 1:1:1;

[0007] Step two: turn on the red light source, and collect the germ rice image Imag_red;

[0008] Step three: turn on the green light source, and collect the germ rice image Imag_green;

[0009] Step four: turn on the blue light source, and collect the germ rice image Imag_blue;

[0010] Step five: respectively perform low threshold filtering processing on the image Imag_red, the image Imag_green and the image Imag_blue;

[0011] Step six: obtain unprocessed rice, and obtain the low threshold filtered images of the unprocessed rice according to steps one to five , image and image , then image , image and images As initial value;

[0012] Step 7: Weighting the low-threshold filtered image Imag_red, image Imag_green, image Imag_blue and the initial value to obtain the relative whiteness value, i.e., the chromaticity of the embryonic rice;

[0013] Step 8: Perform negative feedback control based on relative whiteness value.

[0014] Furthermore, the specific steps of step eight are:

[0015] Step 81: Obtaining the rice germ retention degree according to the relative whiteness value, wherein the relative whiteness value is inversely proportional to the rice germ retention degree;

[0016] Step 82: using the embryo-retained rice image Imag_red, the embryo-retained rice image Imag_green, and the embryo-retained rice image Imag_blue as inputs and the rice germ retention degree as output to train a neural network;

[0017] Step 83: using the trained neural network to identify the rice to be tested, and obtaining the corresponding rice germ retention degree;

[0018] Step 84: Perform negative feedback control based on the rice germ retention rate.

[0019] Furthermore, the low threshold filtering process in step 5 is expressed as:

[0020] Imag_red

[0021] Imag_green

[0022] Imag_blue

[0023] in, Indicates the grayscale of red pixels whose grayscale value is greater than the basic threshold. Indicates the number of red pixels whose grayscale value is greater than the basic threshold. The grayscale value of the green pixel is greater than the basic threshold. Indicates the number of green pixels whose grayscale value is greater than the basic threshold. The grayscale value of the blue pixel is greater than the basic threshold. Indicates the number of blue pixels whose grayscale value is greater than the basic threshold.

[0024] Further, the relative whiteness value is expressed as:

[0025]

[0026] wherein, represents a weighting coefficient.

[0027] Further, the base threshold is 1% of the full scale.

[0028] Further, the neural network is a VGG-19 network, the VGG-19 network is an improved VGG-19 network, and the improvement steps of the improved VGG-19 network are specifically:

[0029] Step one: in the VGG-19 network structure, using a 3*3 convolution kernel to replace the 1, 2, 3, 4 layers of convolution in the 4th convolution module in the VGG-19 network respectively through the extended hollow convolution with the hollow rate value of 1, 2, 5, 7;

[0030] Step two: in the VGG-19 network structure, using a 3*3 convolution kernel to replace the 1, 2, 3, 4 layers of convolution in the 5th convolution module in the VGG-19 network respectively through the extended hollow convolution with the hollow rate value of 1, 2, 5, 7;

[0031] Step three: modify the corresponding pad padding value in the network, that is, when a 3*3 convolution kernel is used to perform convolution operation on a 7*7 feature map, a pixel point with a width of the current rate value is supplemented in the periphery of the feature map to generate a 7*7 size feature map.

[0032] An online automatic detection and control device for embryo rice color, comprising: an industrial camera, a three-primary-color LED light source, a microprocessor, a PLC controller, and a servo cylinder lifting mechanism;

[0033] The three-primary-color LED light source is used for color separation and light distribution of embryo rice.

[0034] The industrial camera is used to collect image information of the embryo rice after color separation and light distribution.

[0035] The microprocessor is used to identify the color of the embryo rice according to the image information of the embryo rice after color separation and light distribution, and to generate a control instruction according to the color of the embryo rice.

[0036] The PLC controller is used to control the cylinder lifting mechanism according to the control instruction.

[0037] Further, the device further comprises an RS-485 communication interface.

[0038] Further, the device further comprises an interactive subsystem, which is used for human-computer interaction.

[0039] Further, the interactive subsystem is a touch display screen.

[0040] The beneficial effects of the present invention are:

[0041] This application primarily addresses the problem of evaluating and detecting rice color. By capturing images of specific detection targets in rice using preset lighting, the system processes and calculates the acquired image data to obtain color evaluation data for the rice, allowing this data to be correlated with the degree of rice processing. The resulting evaluation data is closely correlated and evaluable with the rice's germ retention and hull removal. Compared to traditional human visual observation of rice sample color, this application achieves over 2x the efficiency and more than double the accuracy. BRIEF DESCRIPTION OF THE DRAWINGS

[0042] Figure 1 This is the overall flow chart of this application;

[0043] Figure 2 This is a specific flow chart of the microprocessor for this application;

[0044] Figure 3 This is the hardware structure diagram of this application. DETAILED DESCRIPTION

[0045] It should be noted that, unless there is any conflict, the various embodiments disclosed in this application can be combined with each other.

[0046] Specific implementation method 1: refer to Figure 1 and Figure 2 Specifically describing this embodiment, a method for online automatic detection and control of color of embryo-retained rice described in this embodiment includes the following steps:

[0047] Step 1: Set the RGB brightness values ​​of the three-primary color LED light source to 1:1:1;

[0048] Step 2: Turn on the red light source and collect the embryonic rice image Imag_red;

[0049] Step 3: Turn on the green light source and collect the embryonic rice image Imag_green;

[0050] Step 4: Turn on the blue light source and collect the embryonic rice image Imag_blue;

[0051] Step 5: Perform low threshold filtering on the image Imag_red, the image Imag_green, and the image Imag_blue respectively;

[0052] Step 6: Obtain unprocessed rice and obtain the image of unprocessed rice after low threshold filtering according to steps 1 to 5 ,image and images , then the image ,image and images As initial value;

[0053] Step 7: Weighting the low-threshold filtered image Imag_red, image Imag_green, image Imag_blue and the initial value to obtain a relative whiteness value, i.e., the chromaticity of the embryonic rice;

[0054] Step 8: Perform negative feedback control based on relative whiteness value.

[0055] Specific embodiment 2: This embodiment is a further explanation of specific embodiment 1. The difference between this embodiment and specific embodiment 1 is that the specific steps of step 8 are:

[0056] Step 81: Obtaining the rice germ retention degree according to the relative whiteness value, wherein the relative whiteness value is inversely proportional to the rice germ retention degree;

[0057] Step 82: using the embryo-retained rice image Imag_red, the embryo-retained rice image Imag_green, and the embryo-retained rice image Imag_blue as inputs and the rice germ retention degree as output to train a neural network;

[0058] Step 83: using the trained neural network to identify the rice to be tested, and obtaining the corresponding rice germ retention degree;

[0059] Step 84: Perform negative feedback control based on the rice germ retention rate.

[0060] Specific embodiment three: This embodiment is a further explanation of specific embodiment one. The difference between this embodiment and specific embodiment one is that the low threshold filtering process in step five is expressed as:

[0061] Imag_red

[0062] Imag_green

[0063] Imag_blue

[0064] in, Indicates the grayscale of red pixels whose grayscale value is greater than the basic threshold. Indicates the number of red pixels whose grayscale value is greater than the basic threshold. The grayscale value of the green pixel is greater than the basic threshold. Indicates the number of green pixels whose grayscale value is greater than the basic threshold. The grayscale value of the blue pixel is greater than the basic threshold. Indicates the number of blue pixels whose grayscale value is greater than the basic threshold.

[0065] Specific embodiment 4: This embodiment is a further explanation of specific embodiment 2. The difference between this embodiment and specific embodiment 2 is that the relative whiteness value is expressed as:

[0066]

[0067] in, Represents the weighting coefficient.

[0068] Specific embodiment five: This embodiment is a further explanation of specific embodiment four. The difference between this embodiment and specific embodiment four is that the basic threshold is 1% of the full scale.

[0069] Specific embodiment 6: This embodiment is a further explanation of specific embodiment 1. The difference between this embodiment and specific embodiment 1 is that the neural network is a VGG-19 network, and the VGG-19 network is an improved VGG-19 network. The steps of improving the improved VGG-19 network are specifically as follows:

[0070] Step 1: In the VGG-19 network structure, use the 3*3 convolution kernel with the dilated convolution rate values ​​of 1, 2, 5, and 7 to replace the 1st, 2nd, 3rd, and 4th convolution layers in the 4th convolution module of the VGG-19 network respectively;

[0071] Step 2: In the VGG-19 network structure, use the 3*3 convolution kernel with the expanded dilated convolution rate of 1, 2, 5, and 7 to replace the 1st, 2nd, 3rd, and 4th convolution layers in the 5th convolution module of the VGG-19 network respectively;

[0072] Step 3: Modify the corresponding padding value in the network. That is, when performing a convolution operation on a 7*7 feature map with a 3*3 convolution kernel, add pixels with a width equal to the current rate value on the periphery of the feature map to generate a 7*7 feature map.

[0073] Specific implementation method seven: refer to Figure 3 Specifically describing this embodiment, an online automatic detection and control device for color of embryo-retained rice described in this embodiment includes: an industrial camera, a three-primary-color LED light source, a microprocessor, a PLC controller, and a servo electric cylinder lifting mechanism;

[0074] The three-primary-color LED light source is used to perform color separation and light distribution on the embryo-retained rice;

[0075] The industrial camera is used to collect image information of the remaining embryo rice after color separation and light distribution;

[0076] The microprocessor is configured to identify the germ rice chroma according to the image information of the germ rice after the color separation and light distribution, and generate a control instruction according to the germ rice chroma;

[0077] The PLC controller is configured to control the electric cylinder lifting mechanism according to the control instruction.

[0078] Specific embodiment eight: this embodiment is a further description of specific embodiment seven, and the difference between this embodiment and specific embodiment seven is that the device further comprises an RS-485 communication interface.

[0079] Specific embodiment nine: this embodiment is a further description of specific embodiment eight, and the difference between this embodiment and specific embodiment eight is that the device further comprises an interactive subsystem, and the interactive subsystem is configured to realize human-computer interaction.

[0080] Specific embodiment ten: this embodiment is a further description of specific embodiment nine, and the difference between this embodiment and specific embodiment nine is that the interactive subsystem is a touch display screen.

[0081] The application adopts a machine vision system to collect real-time images of germ rice in the processing of rice, and realizes online detection of rice according to the invented germ rice chroma information processing and evaluation algorithm. The system adopts a self-defined controllable light source to provide chroma preset for image collection, and adopts a color separation and light distribution collection method to complete digital collection of online images of rice. According to image data obtained by red, green and blue three light distributions under three primary color light distributions, a color separation and weighting algorithm is adopted to calculate the color evaluation index of rice. The final calculation index is the relative whiteness value of rice. Finally, the system applies the relative whiteness value provided by the algorithm to production control. The relative whiteness value is inversely proportional to the germ retention of rice.

[0082] The rice sample flowing out of the discharge port of the rice mill is sampled by a device composed of an industrial camera, an LED light source, a glass window, etc. The sampled rice sample information is loaded into a PLC programmable controller through algorithm optimization processing after being compared by an embedded microprocessor and a germ rice database, and the PLC controls the servo electric cylinder to rise, which is used to adjust the gap between the rice milling knife group and the rice milling sand belt, i.e. the height distance of the transverse rolling of the rice grains. Thus, the color depth of the germ rice, i.e. the chroma, is adjusted.

[0083] Since the system is a negative feedback system, if the color of the rice sample at the discharge port is deeper than the set color, it means that the gap is large, and the feedback signal can control the servo electric cylinder to descend (in microns) through the microprocessor and the PLC, so as to achieve the set chroma of the germ rice. On the contrary, if the color of the rice sample at the discharge port is light, it means that the gap is small, and the sampling signal can control the servo electric cylinder to rise (in microns) through the microprocessor and the PLC, so as to automatically achieve the requirement of the set chroma.

[0084] The man-machine interface composed of the PLC and the touch screen can set various parameters according to requirements. Meanwhile, the man-machine interface can be designed to be friendly, dynamically display the running state of the machine, self-diagnose, automatically alarm and the like.

[0085] The application solves the problems of low efficiency, poor accuracy, high broken rice rate and poor rice quality caused by human eye observation in the prior art, and solves the defects in the prior art, improves the processing quality and production efficiency of the rice with germ, and realizes automation and intelligentization.

[0086] It should be noted that the specific implementation is only an explanation and description of the technical scheme of the application, and cannot limit the protection scope. Any partial change according to the claims and description of the application shall still fall within the protection scope of the application.

Claims

1. A method for automatically detecting and controlling the color of embryo-retained rice, characterized in that The following steps are involved: Step 1: Set the RGB brightness values ​​of the three-primary color LED light source to 1:1:1; Step 2: Turn on the red light source and collect the embryonic rice image Imag_red; Step 3: Turn on the green light source and collect the embryonic rice image Imag_green; Step 4: Turn on the blue light source and collect the embryonic rice image Imag_blue; Step 5: Perform low threshold filtering on the image Imag_red, the image Imag_green, and the image Imag_blue respectively; Step 6: Obtain unprocessed rice and obtain the image of unprocessed rice after low threshold filtering according to steps 1 to 5 ,image and images , then the image ,image and images As initial value; Step 7: Weighting the low-threshold filtered image Imag_red, image Imag_green, image Imag_blue and the initial value to obtain a relative whiteness value, i.e., the chromaticity of the embryonic rice; Step 8: Perform negative feedback control based on relative whiteness value; The low threshold filtering process in step 5 is expressed as: Imag_red Imag_green Imag_blue in, Indicates the grayscale of red pixels whose grayscale value is greater than the basic threshold. Indicates the number of red pixels whose grayscale value is greater than the basic threshold. The grayscale value of the green pixel is greater than the basic threshold. Indicates the number of green pixels whose grayscale value is greater than the basic threshold. The grayscale value of the blue pixel is greater than the basic threshold. Indicates the number of blue pixels whose grayscale value is greater than the basic threshold; The relative whiteness value is expressed as: in, Represents the weighting coefficient.

2. The method for automatically detecting and controlling the color of embryo-retained rice according to claim 1, wherein The specific steps of step eight are: Step 81: Obtaining the rice germ retention degree according to the relative whiteness value, wherein the relative whiteness value is inversely proportional to the rice germ retention degree; Step 82: using the embryo-retained rice image Imag_red, the embryo-retained rice image Imag_green, and the embryo-retained rice image Imag_blue as inputs and the rice germ retention degree as output to train a neural network; Step 83: using the trained neural network to identify the rice to be tested, and obtaining the corresponding rice germ retention degree; Step 84: Perform negative feedback control based on the rice germ retention rate.

3. The method for automatically detecting and controlling the color of embryo-retained rice according to claim 1, wherein The basic threshold is 1% of the full scale.

4. The method for automatically detecting and controlling the color of embryo-retained rice according to claim 2, wherein The neural network is a VGG-19 network, and the VGG-19 network is an improved VGG-19 network. The steps of improving the improved VGG-19 network are specifically as follows: Step 1: In the VGG-19 network structure, use the 3*3 convolution kernel after expansion with the dilated convolution rate value of 1, 2, 5, and 7 to replace the 1st, 2nd, 3rd, and 4th convolution layers in the 4th convolution module of the VGG-19 network respectively; Step 2: In the VGG-19 network structure, use the 3*3 convolution kernel after expansion with the dilation rate value of 1, 2, 5, and 7 to replace the 1st, 2nd, 3rd, and 4th convolution layers in the 5th convolution module of the VGG-19 network respectively; Step 3: Modify the corresponding padding value in the network. That is, when performing a convolution operation on a 7*7 feature map with a 3*3 convolution kernel, add pixels with a width equal to the current rate value on the periphery of the feature map to generate a 7*7 feature map.

5. A control device for implementing the control method according to any one of claims 1 to 4, characterized in that include: Industrial camera, three-primary-color LED light source, microprocessor, PLC controller and servo electric cylinder lifting mechanism; The three-primary-color LED light source is used to perform color separation and light distribution on the embryo-retained rice; The industrial camera is used to collect image information of the remaining embryo rice after color separation and light distribution; The microprocessor is used to identify the chromaticity of the embryonic rice according to the image information of the embryonic rice after color separation and light distribution, and generate a control instruction according to the chromaticity of the embryonic rice; The PLC controller is used to control the electric cylinder lifting mechanism according to control instructions.

6. The device for automatically detecting and controlling the color of embryo-retained rice according to claim 5, characterized in that The device also includes an RS-485 communication interface.

7. The device for automatically detecting and controlling the color of embryo-retained rice according to claim 6, characterized in that The device further comprises an interaction subsystem, and the interaction subsystem is used for human-computer interaction.

8. The device for automatically detecting and controlling the color of embryo-retained rice according to claim 7, characterized in that The interactive subsystem is a touch display screen.

Images