Grain-by-grain detection method and grain-by-grain selection device for pre-embryo integrity of brown rice used as rice seeds

Abstract

The invention relates to the field of rice seed selection, in particular to a grain-by-grain detection method and a grain-by-grain seed selection device for the integrity of pre-embryos of brown rice used as rice seeds, and aims to solve the problems that in the prior art, the integrity of the brown rice embryos cannot be accurately identified at low cost, qualified and unqualified brown rice cannot be separated one by one, so that the survival rate of selected seeds is low, and the seed selection efficiency is low. The seed selection cost is high in the seed selection process of the brown rice. According to the method, algorithm recognition is carried out on the integrity of the brown rice embryos, and the brown rice with complete embryos and the brown rice with incomplete embryos are accurately separated. The brown rice without the rice hulls avoids the generation of bacteria, and further does not need to be disinfected. A large amount of preparation time for seed drying and disinfection procedures is reduced, germination is directly carried out, the germination period is greatly shortened due to the absence of rice hulls, and the cost is low. The rice seed selection device is used in the seed selection process before rice breeding.

Description

Technical Field

[0001] The invention relates to the field of rice seed selection, and in particular to a grain-by-grain detection method for embryo integrity of brown rice before it is used as rice seeds and a grain-by-grain seed selection device, which are used for seed selection before seed breeding. Background Art

[0002] Rice is one of the main food crops in the world and the main agricultural crop in my country. Its annual output and planting area rank first in the world. Among the countries that grow rice in the world, our country has relatively high output. As of March 2024, my country's total rice output accounts for 30% of the global total, so it has a very important position in the agricultural economy.

[0003] Because the outer shell of rice is relatively hard, the rice seeds need to be processed before planting in order to increase its germination rate and emergence rate. First, the seeds need to be exposed to the sun, which can kill the bacteria on the surface of the rice skin. Then, after diluting the disinfectant solution, the rice seeds are placed in the solution for soaking. The soaking time must be precise at this time, because the disinfectant solution itself is irritating. If the time is too long, it is easy to damage the seeds. Generally, the soaking time is controlled within five minutes.

[0004] Then soak it in germination powder for about twenty-four hours. Wait until the rice grains are all expanded before germination can be carried out. During germination, the ambient temperature needs to be controlled well. It should be placed in a relatively warm and humid environment. This can improve the efficiency of germination. Wait until about 80% of the seedlings have all emerged as white sprouts before sowing. This can increase the germination rate of rice seedlings and increase the activity of seeds.

[0005] According to the existing rice seed selection process, the following technical problems exist in the seed selection process:

[0006] 1. Since rice seeds need to go through processes such as drying, disinfection, soaking and germination, there is a problem of a long seed selection cycle.

[0007] 2. In the seed selection process, if rice seeds that cannot germinate are encountered, direct economic losses will be caused. This seed selection method has the problem of high seed selection costs.

[0008] 3. Authorization announcement number is CN111802242B, and the patent name is a hybrid rice seedling raising method utilizing brown rice seedlings. In order to obtain brown rice with high germination rate, when screening brown rice, the technical means adopted is to utilize photoelectric colorimeter to screen brown rice. Although it can remove brown rice grains that are incomplete, mildewed or other bad to a certain extent, the use of photoelectric colorimeter requires a relatively large expenditure, and a photoelectric colorimeter generally costs 20,000 to 30,000 yuan. Therefore, it has the problem of high seed selection cost, and the photoelectric colorimeter utilizes photoelectric conversion elements such as photocells or photoelectric tubes as detectors to measure the intensity of transmitted light after passing through a colored solution, thereby deriving a method for the content of a measured substance (for example, the measured substance is the embryo on the brown rice grain, and it can measure the content of the embryo), and it is not used for the detection of embryo integrity.

[0009] 4. Whether the brown rice can germinate, take root and become seedlings after selection is closely related to whether the embryo on the brown rice is intact. Brown rice with intact embryos has better vitality, and its embryo germination and root growth momentum are good. However, it is impossible to identify whether the embryo contained in rice is intact because the rice husk is opaque. In addition, in the process of husking rice to become brown rice, due to improper operation, the embryo may be damaged to varying degrees. Therefore, not all brown rice can be successfully bred, so the identification of the integrity of the embryo is crucial.

[0010] The patent application numbered 202211456778.4 filed by Harbin Institute of Technology, although its purpose is to identify the integrity of the embryo, the solution only discloses that "it can distinguish whether the rice has embryos and calculate the embryo retention rate", and it does not disclose how to calculate the embryo retention rate, especially whether the embryo retention rate refers to how many rice grains still have embryos in all the rice grains, or how many embryos are retained in a germ rice, and there is no clear explanation. According to the records of its background technology, it is essentially used as edible rice to detect embryos. Edible rice only considers whether it has embryos. Therefore, its embryo retention rate should refer to how many rice grains still have embryos in all the rice grains. For this reason, it does not need to pay attention to whether the embryos left can germinate to make rice seeds. It does not and will not clearly explain the integrity detection of the "embryo" itself, that is, it does not need to consider whether the embryo can germinate, and naturally will not consider how many embryos should be retained for a germ rice to be considered intact. Therefore, its related technology for embryo integrity is not fully disclosed. What is more critical is: if the problem of embryo integrity is not taken into consideration, since the scheme uses the reflectivity of reflected light to achieve recognition, even if the scheme is used to identify embryos in the process of processing embryo-retained rice, the reflected light will be greatly affected by the dark environment of other enclosed spaces. Although it also discloses "appropriately increase the exposure time of the camera to obtain an image of appropriate brightness", different exposure times in dark environments will also lead to very different images collected in different situations, at different times, and under different working conditions, thus seriously affecting the recognition effect. The need for brown rice to be used as a pre-embryo integrity test for rice seeds is based on "seed selection equipment / machinery". In this case, the seed selection workspace is basically a "closed or semi-closed" environment, which creates a relatively dark environment. In order to ensure image acquisition, a light source that provides lighting conditions needs to be set. If the light source is damaged and replaced, the relatively dark environment will undergo major changes, which will affect the accuracy of reflected light detection and seriously affect 202211456778.4, a type of detection scheme based on reflected light, which may greatly affect the accuracy of rice seed pre-embryo integrity detection, leading to inaccurate seed selection, affecting the germination rate, and possibly leading to a reduction in seed yield after planting.

[0011] Although Harbin Engineering University's patent application with application number 202110284189.1 can also be used to identify the integrity of the germ, it is implemented based on the SVM method. Supervised machine learning methods such as SVM require training of recognition models. In the process of training the recognition model, a large number of labeled samples need to be prepared, that is, a large number of germ rice images need to be annotated one by one. The marking of ordinary image samples requires a lot of time and workload costs, especially the number of samples directly affects the effect of the model. To achieve good recognition results, a large number of samples need to be marked. For example, according to eight thousand samples, each sample needs to be marked one by one using tools, which will take more than half a month of time and workload; what's more, for very small images such as rice grains, even if they are enlarged for marking, the marking time and marking workload will be greatly increased. More importantly, the morphology of brown rice grains of different varieties of rice is very different. Even for the same variety of rice, the appearance of brown rice grains produced each year will be different, and the appearance of brown rice grains produced in different regions is also very different. This will lead to poor applicability of machine learning. At least a large number of brown rice grains of different types of rice, rice from different regions, and even rice from different years need to be marked, which will cause the time cost and workload cost of marking to increase by several times or even dozens of times, which is difficult for general companies to accept; at the same time, for the marked samples, the cost of building and / or training the supervised machine learning model is also very huge, which is unaffordable for general development companies. In addition, the hardware equipment running the machine learning model generally requires the use of hardware with relatively high performance, which will also increase the cost of the equipment. This restricts the research and development process of the seed selection equipment and the final market penetration and share of the equipment. Therefore, there is a demand in the market to develop low-cost, high-germination rate seed selection equipment.

[0012] In summary, the existing processing method of rice with embryo retained cannot accurately and cost-effectively identify the integrity of brown rice embryos, and cannot separate qualified and unqualified brown rice grain by grain, so the germination results of the selected seeds are not very good, which in turn affects the yield, and the problem of high seed selection cost in the brown rice selection process exists. Summary of the invention

[0013] The invention aims to solve the problems that the existing processing method of rice with embryos cannot accurately and cost-effectively identify the integrity of brown rice embryos, and cannot separate qualified and unqualified brown rice grain by grain, so the germination result of selected seeds is not very good, thereby affecting the yield, and the high seed selection cost in the brown rice seed selection process, and further provide a grain-by-grain detection method for the integrity of the embryos of brown rice before it is used as rice seeds and a grain-by-grain seed selection device.

[0014] The technical solution of the present invention is:

[0015] The present invention provides a method for detecting the integrity of embryos of brown rice before it is used as rice seeds, which comprises the following steps:

[0016] S1, obtaining a color image of brown rice grains when a light source is transmitted through the brown rice grains;

[0017] S2, preprocessing the color brown rice grain image to obtain a preprocessed color image;

[0018] S3, converting the preprocessed color image into a grayscale image;

[0019] S4, performing light correction on the grayscale image;

[0020] S5, based on the grayscale image after light correction, based on the first threshold, utilize threshold segmentation method to segment the non-brown rice grain region and the brown rice grain region of each brown rice; For each brown rice grain region, carry out subsequent processing to the brown rice grain region one by one;

[0021] S6. For the brown rice grain region in the grayscale image, based on the second threshold, using a threshold segmentation method to segment the embryo region in the brown rice grain region, and obtaining an outline of the embryo region, which is recorded as a first outline;

[0022] S7, mapping the brown rice grain area in the grayscale image back to the color image of step S2, and extracting the brown rice grain area in the filtered color image;

[0023] Then, the color image of the brown rice grain area is converted into the HSV space, and the V component image is extracted; for the V component image, the embryo region of the brown rice grain area in the V component image is segmented by setting a third threshold and using a threshold segmentation method, and the contour of the embryo region is obtained, which is recorded as the second contour;

[0024] S8, mapping the second contour to the brown rice grain area corresponding to step S6, thereby obtaining the mapped contour of the second contour and the first contour on the image, and extracting contour points based on the mapped contours of the first contour and the second contour: extracting corresponding points from the two contours, and then interpolating these points to generate new contour points, thereby obtaining a more accurate fused contour;

[0025] S9, based on the fused contour in step S8, convex hull detection is performed, and if the convex hull detection passes, it is determined as the initial inspection embryo area, otherwise it is considered that the embryo is incomplete;

[0026] S10. Determine the integrity of the embryo in the initial embryo inspection area.

[0027] Furthermore, the process of preprocessing the color brown rice grain image in step S2 includes Gaussian filtering.

[0028] Furthermore, in the process of judging the integrity of the embryo in the initial inspection embryo area in step S10, the integrity of the embryo is judged based on the area characteristics of the initial inspection embryo area.

[0029] Furthermore, in the process of judging the integrity of the embryo based on the area characteristics of the initial inspection embryo region, when the area of ​​the embryo region is greater than a set area threshold, the embryo is judged to be complete, otherwise the embryo is considered to be incomplete.

[0030] or,

[0031] The brown rice grain region image corresponding to the fusion contour determined in step S8 is extracted as a whole, and stretched and transformed to a specified size; then, the area of ​​the embryo region is calculated based on the fusion contour in the stretched transformed image within the specified size range. If it is greater than the set area threshold, the embryo is judged to be complete, otherwise, the embryo is considered to be incomplete.

[0032] or,

[0033] For the initial inspection embryo region, the area of ​​the embryo region is calculated; at the same time, the area of ​​the brown rice grain region in step S5 is calculated. If the ratio of the area of ​​the embryo region to the area of ​​the brown rice grain region is greater than the ratio threshold, the embryo is judged to be complete, otherwise it is considered that the embryo is incomplete.

[0034] The present invention also provides a seed selection device for detecting the integrity of the embryo before the brown rice is used as rice seeds, which comprises a shell, a rice dropping bin, a rice dropping plate and a tormi board, wherein the tormi board is mounted on the shell, the rice dropping plate is rotatably mounted on the tormi board, the rice dropping plate is provided with a plurality of rice dropping grooves in a circular array in the circumferential direction, and the rice dropping bin is mounted on the rice dropping board; the device also comprises a cover body, a camera, a light belt and a flap, wherein the light belt is embedded under the tormi board, the flap can be opened and mounted on the tormi board, a flap mounting hole is provided on the tormi board at a position corresponding to the flap, a separation hole is also provided on the tormi board, the camera is mounted above the rice dropping plate through the cover body, and the camera is located above the light belt, wherein a camera module is integrated in the camera, and a grain-by-grain detection method for detecting the integrity of the embryo before the brown rice is used as rice seeds is operated in the camera module.

[0035] Furthermore, the rice falling bin, the light strip, the flap and the separation hole are arranged at equal angular intervals along the circumferential direction.

[0036] Furthermore, it also includes a flap drive assembly, which includes a motor, a guide roller, a rope and two bearing seats. The motor and the two relatively arranged bearing seats are installed on the rear end surface of the tommy plate, the two shaft ends of the guide roller are rotatably installed on the two bearing seats, and the output shaft of the motor is connected to one end of the guide roller, one end of the rope is fixed on the guide roller, and the other end of the rope is wound around the guide roller and the rope end is fixedly connected to the back side of the flap.

[0037] Furthermore, it also includes a rotating shaft and a spring. The flap is rotatably mounted on the tommy plate via the rotating shaft. The spring is sleeved on the rotating shaft and abuts against the flap and the flap mounting hole on the tommy plate.

[0038] Furthermore, it also includes a rice dropping bin, which is installed on the rice dropping board.

[0039] Compared with the prior art, the present invention has the following effects:

[0040] 1. Compared with the existing brown rice seed selection, the brown rice seeds selected by the grain-by-grain detection method and the grain-by-grain seed selection device of the present invention have the advantages of better germination results:

[0041] Firstly, the present invention recognizes the image of brown rice transmitted by the light source, which can greatly reduce or even avoid the influence of the light environment change in a dark environment, thereby ensuring the accuracy of the detection.

[0042] The present invention adopts the fusion of grayscale image and V component image for detection, reduces the error caused by the influence of illumination, grayscale image itself and other problems, further improves the detection effect, and at the same time, the present invention integrates V component image on the basis of grayscale image, which can ensure the stability of the relative difference between endosperm and embryo under different environments, improves the robustness of detection, and adopts V component image for fusion to determine the embryo area, which can greatly reduce the aforementioned influence and improve the accuracy of embryo area determination. The cover body is buckled on the frame, and during the shooting process, it plays the function of shielding the external light source, forming a dark shooting environment, avoiding the influence of other light except the light source, and can improve the detection effect to a certain extent. On this basis, the scheme of the present invention processes the transmission image and considers the fusion processing of V component image, which can greatly reduce or even avoid the influence of light environment changes caused by the replacement of light source in dark environment, always ensure the detection accuracy, and reduce the parameter requirements for the shooting head, thereby reducing the cost. In addition, since the V component image can ensure the stability of the relative difference between the endosperm and the embryo, it is also more conducive to displaying the state of the embryo, and it is easier to determine the degree of embryo damage later, which is conducive to the detection of the convex hull, improves the recognition effect of the convex hull, and ensures that the area detection is combined on the basis of the convex hull detection to ultimately ensure the detection effect of whether the embryo is complete. The detection method of the present invention does not require training of the model, which saves the training cost itself, and also avoids a large number of labeled samples required for the training process, that is, the present invention does not need to mark and produce labeled samples, so it also saves the cost of producing training samples, thereby reducing the development cost.

[0043] The present invention identifies the embryo integrity of brown rice before breeding and accurately separates the brown rice with intact embryos from the brown rice with incomplete embryos. This process omits the traditional rice seed drying and disinfection steps, so that manufacturers can directly use qualified brown rice for soaking and other steps.

[0044] Secondly, the reason why the germination rate of the embryo end is high is that the present invention uses qualified brown rice for direct germination, and the germination rate of brown rice with complete embryos is bound to be high.

[0045] Again, the germination rate of the intact embryo brown rice used in the present invention in the germination experiments was above 98%, and the average sprout length was 40 mm-50 mm.

[0046] Therefore, the present invention adopts qualified brown rice for germination breeding, which can effectively improve the germination effect.

[0047] 2. The germination period of the present invention is short:

[0048] Since the seed selection of the present invention firstly performs algorithm recognition on the embryo integrity of brown rice, and accurately separates the brown rice with complete embryo from the brown rice with incomplete embryo. The brown rice without rice husk avoids the generation of bacteria, and no disinfection is required. The preparation time of the drying and disinfection process is greatly reduced, and the germination can be directly carried out. Because there is no rice husk, the germination period is greatly shortened. BRIEF DESCRIPTION OF THE DRAWINGS

[0049] Figure 1 The image of the brown rice grains within the collection field of view when the image is collected on the rice falling plate 2 in the seed selection device of the present invention;

[0050] Figure 2 To obtain the brown rice grain area image by segmentation algorithm;

[0051] Figure 3 is the image of the brown rice grain area in the extracted filtered color image;

[0052] Figure 4 This is an enlarged schematic diagram of the structure of the rice embryo. The functions of the various tissue structures of the embryo in the figure are as follows: (1) The incomplete leaflets, first leaflet, second leaf primordium, and stem growth point in the figure are collectively referred to as the embryo. The integrity of the embryo is a necessary condition for the development of the stem and leaves of the seedling, and it provides genes for the stem and leaves of the seedling. (2) The radicle sheath, radicle cap, and radicle growth point in the figure are collectively referred to as the radicle. The integrity of the radicle is a necessary condition for the development of the seedling root, and it provides genes for the seedling root. (3) The embryonic shield in the figure transports nutrients to the embryo and radicle. (4) The embryonic sheath in the figure protects the embryo and plays a role in helping the buds and leaves to arch upward and break through the soil.

[0053] Figure 5 Schematic diagram of embryo status distinction; the "not full" state shown in Figure (a) is the "embryo damaged" state, and the "full" state shown in Figure (b) is the "embryo complete" state;

[0054] Figure 6It is a schematic diagram of the overall structure and appearance of the seed selection device of the present invention;

[0055] Figure 7 yes Figure 6 A longitudinal section view of

[0056] Figure 8 It is an axonometric view of Tomi Plate 3;

[0057] Fig. 9 yes Figure 8 Bottom view of

[0058] Fig.10 yes Fig. 9 A partial enlarged view at point A.

[0059] Fig.11 This is a photo of a germination experiment of brown rice with qualified embryos after selection by the grain-by-grain selection device of the present invention.

[0060] Fig.12 is with Fig.11 Photo of brown rice that failed the embryo integrity test undergoing germination experiments under the same experimental conditions. DETAILED DESCRIPTION Specific implementation method one:

[0062] This embodiment is a method for detecting the integrity of embryos of brown rice before it is used as rice seeds, comprising the following steps:

[0063] S1. Image acquisition:

[0064] Under the condition of uniform background light illumination, the light source is transmitted through the brown rice grains, and the image is acquired using the OpenMV4 H7 Plus camera to obtain the image of the brown rice grains when the light source is transmitted through the brown rice grains. The image of the brown rice grains is an RGB color image. In the process of acquiring images, it is necessary to ensure that clear image data is obtained.

[0065] Since the image is obtained by transmitting the light source through the brown rice grains, the subsequent processing is performed on the transmission image. This method itself can greatly reduce or even avoid the influence of light environment changes caused by light source replacement in a dark environment, and always ensure the accuracy of detection; in addition, because the image is obtained by transmitting the light source through the brown rice grains, in the subsequent detection process, in addition to using the grayscale image for segmentation, the characteristics of the transmission image can also be used to jointly determine the embryo area, thereby reducing the errors caused by the influence of light, the grayscale image itself and other problems, and improving the detection effect; more importantly, this method is basically not affected by factors such as the variety, region, moisture content and other factors of the rice corresponding to the brown rice, and greatly ensures the consistency of the recognition effect of different brown rice used as rice seeds.

[0066] S2. Image preprocessing:

[0067] The collected brown rice grain image is subjected to Gaussian filtering to obtain a filtered color image.

[0068] It should be noted that: in this embodiment, Gaussian filtering is used for processing, which can retain more image details and is conducive to subsequent image segmentation processing. In fact, this process can also be processed by other filtering methods.

[0069] S3. Image grayscale processing:

[0070] Convert the image to grayscale, a process that helps highlight the area of ​​the brown rice germ.

[0071] S4. Light correction:

[0072] In order to deal with the problem of uneven light in the environment, the histogram equalization method is used to correct the light of the image and balance the image brightness, making the features of the brown rice germ area more prominent.

[0073] S5. Segmentation of brown rice region in grayscale image:

[0074] Based on the grayscale image after light correction, a first threshold is set and a threshold segmentation method is used to segment each grain of brown rice to obtain a non-brown rice grain area (background) and a brown rice grain area (foreground).

[0075] It should be noted that: due to the camera setting position and setting mode in step S1, the collected brown rice grain image contains multiple brown rice grains. In this embodiment, based on the illumination range of the light source and the setting of the camera, the image field of view is generally 4 rice troughs, such as Figure 1 As shown in the figure, each image generally contains at most 4 grains of brown rice. After segmenting each grain of brown rice to obtain the non-brown rice grain area (background) and the brown rice grain area (foreground), only 1-2 brown rice grain areas in the middle of the image field of view are selected (foreground as shown in the figure). Figure 2 In fact, if the setting position and setting mode of the camera in step S1 (depending on the structure and spatial setting of the seed selection device) make the collected brown rice grain image contain only one brown rice grain, then it can be directly processed, which can save the amount of calculation to a certain extent, but it should be noted that this will affect the image filtering effect and the effect of light correction.

[0076] S6. For the brown rice grain region in the grayscale image, a second threshold is set and a threshold segmentation method is used to segment the brown rice grain region to obtain an embryo region, and an outline of the embryo region is obtained, which is recorded as a first outline.

[0077] S7, V component image segmentation:

[0078] The brown rice grain region in the grayscale image is mapped back to the filtered color image, and the brown rice grain region in the filtered color image is extracted, such as Figure 3 As shown;

[0079] Then, the color image of the brown rice grain area is converted into the HSV space, and the V component image is extracted. For the V component image, the third threshold is set and the threshold segmentation method is used to segment the embryo area of ​​the brown rice grain area in the V component image, and the outline of the embryo area is obtained, which is recorded as the second outline.

[0080] S8, Contour Fusion:

[0081] The second contour is mapped to the brown rice grain area image corresponding to step S6, so as to obtain the mapped contour of the second contour and the first contour on the image, and the contour points are extracted based on the mapped contours of the first contour and the second contour: corresponding points are extracted from the two contours, and then these points are interpolated to generate new contour points, so as to obtain a more accurate fused contour (the contour corresponding to the embryo area).

[0082] It should be noted that: the present invention uses the image obtained by transmitting the brown rice grains through the light source for detection. Even in this case, the influence of the lighting environment conditions can be reduced or avoided. However, due to the characteristics of the grayscale image itself, segmentation errors can be caused to a certain extent, especially for the brown rice obtained from different varieties and different quality rice. There are relatively large differences in the brown rice itself, which will also lead to deviations in the determination of the obtained embryo area. The present invention is based on the analysis and research of the image, and incorporates the V component image. The image can ensure the stability of the relative difference between the endosperm and the embryo under different environments, improves the robustness of the detection, and uses the V component image to fuse and determine the embryo area, which can greatly reduce the aforementioned influence and improve the accuracy of the embryo area determination. In addition, it has been found through research that since the V component image can ensure the stability of the relative difference between the endosperm and the embryo, this is also more conducive to displaying the state of the embryo, and it is easier to determine the degree of embryo damage later.

[0083] S9, Fusion Convex Hull Detection:

[0084] Based on the fused contour in step S8, convex hull detection is performed. If the convex hull detection passes, it is determined as the initial inspection embryo area, otherwise the embryo is considered incomplete.

[0085] After research, it was found that under normal circumstances, the embryo should be in a "protruding" state on the outline of brown rice. Due to the shape caused by the composition of the embryo itself and the influence of the strength of its combination with the endosperm, once the embryo is worn, there is a high probability that a "gap" will appear. Since brown rice can only be obtained by "milling" rice, but the degree of "milling" is relatively light, even after slight "milling", the "brown rice" will tend to become "elliptical", that is, the shape changes towards a "convex hull". In other words, if there is still a "concave" shape after milling, it is very likely that the embryo has been completely ground off, or the wear is relatively serious, so the following is used. Figure 5 The forms shown are used to distinguish the states of the embryos, Figure 5 The "unsatisfied" state shown in (a) is the "embryo damage" state. Figure 5 The "full" form shown in (b) is the "embryo complete" state. It should be noted that the "embryo complete" state does not mean that the embryo is 100% complete, but is only complete in morphology. In order to ensure that the embryo can germinate and develop into breeding, the present invention also needs to ensure the degree of integrity of the embryo, which is the basic requirement for the complete convex hull detection. Figure 4 This is an enlarged schematic diagram of the structure of a rice embryo. From the figure, it can be seen that if brown rice is to germinate normally, the integrity of the "initial state" of the embryo should be ensured as much as possible, or all the structures of the embryo should be preserved as much as possible. Therefore, the present invention continues to detect the integrity of the preserved state.

[0086] S10. Determination of embryo integrity:

[0087] For the initial embryo inspection area, calculate its area and other characteristics to determine the integrity of the embryo:

[0088] Solution 1: If the area of ​​the embryo region is greater than the set area threshold, the embryo is considered complete, otherwise it is considered incomplete.

[0089] It should be noted that the image processing method in the aforementioned steps of the present invention can reduce or even be substantially unaffected by factors such as rice variety, region, and moisture content, but the embryo size is related to factors such as the variety. If one wants to improve the overall detection effect and ensure the effect of embryo germination, one needs to consider factors such as the variety and origin of brown rice, as well as external parameters of the image acquisition device (such as the distance between the image acquisition device and the brown rice) to determine the area threshold. If one wants to make a more accurate judgment, one also needs to consider factors such as the degree of brown rice grinding each time.

[0090] In the process of judging the integrity of the embryo, in order to eliminate the influence of different varieties and origin factors of brown rice as much as possible, the brown rice grain area image corresponding to the fusion profile determined in step S8 can be extracted as a whole, stretched (can be negative stretching) to transform the specified size (fixed length and width), and then the area of ​​the embryo area is calculated based on the fusion profile in the stretched transformed image of the specified size range (the fusion profile is the fusion profile after the stretch transformation of the image), and the embryo is judged to be complete if it is greater than the set area threshold, otherwise it is considered that the embryo is incomplete. It should also be supplemented that: at this time, the area threshold is also determined based on the image after the stretch transformation, and can be determined according to the actual situation of the image after the stretch transformation.

[0091] Solution 2: If the influence of the external parameters of the image acquisition device is to be eliminated, the area ratio and the ratio threshold can also be used to determine, that is, the area of ​​the embryo region is calculated, and the area of ​​the brown rice grain region in step S5 is calculated at the same time. If the ratio of the area of ​​the embryo region to the area of ​​the brown rice grain region is greater than the ratio threshold, the embryo is judged to be complete, otherwise it is considered that the embryo is incomplete.

[0092] So far, this embodiment has completed the detection of embryo integrity of brown rice used for rice seeds. Considering that visualization processing and seed selection (seed selection for brown rice with intact embryos) and other processing are required in the later stage, the corresponding brown rice needs to be selected at the same time. Therefore, it may be necessary to extract other characteristics of brown rice, which will also involve embryo position detection. Therefore, the detection method of this embodiment can also include the step of embryo position detection:

[0093] According to the morphological characteristics of brown rice, the rice grain area is divided into two parts in the length direction, and the distribution of pixels where the embryo is located is counted. If the embryo is mainly concentrated in the upper part, it is determined as the upper end; if it is concentrated in the lower part, it is determined as the lower end.

[0094] Mark the location of the brown rice embryo in the image to ensure visualization of the recognition results.

[0095] Completeness label: Based on the completeness judgment result, a "complete" or "incomplete" label is displayed in the detection box.

[0096] Position label: Based on the position determination result, mark “top” or “bottom” next to the box.

[0097] Specific implementation method 2: Combination Figures 6 to 10The present embodiment is described. The present embodiment is a grain-by-grain seed selection device, which includes a shell 1, a rice-falling bin 15, a rice-falling plate 2 and a tormi board 3. The tormi board 3 is mounted on the shell 1, and the rice-falling plate 2 is rotatably mounted on the tormi board 3. The rice-falling plate 2 is provided with a plurality of rice-falling grooves 4 in a circular array in the circumferential direction, and the rice-falling bin 15 is mounted on the rice-falling plate 2; it also includes a cover body 5, a camera 6, a light strip 7 and a flap 8. The light strip 7 is embedded under the tormi board 3, the flap 8 can be opened and mounted on the tormi board 3, and a flap mounting hole 9 is provided on the tormi board 3 at a position corresponding to the flap 8, and a separation hole 16 is also provided on the tormi board 3. The camera 6 is mounted above the rice-falling plate 2 through the cover body 5, and the camera 6 is located above the light strip 7, wherein a camera module is integrated in the camera 6, and the camera module contains brown rice for a grain-by-grain detection method of the integrity of the pre-embryo of rice seeds.

[0098] With such arrangement, the present embodiment provides a device for screening out brown rice with intact embryos one by one, and the screening method mainly uses the camera 6 to take a picture of the light-transmitting image of the brown rice in the rice falling trough 4, and an algorithm and a control program are implanted in the chip, and the algorithm is used to accurately identify the brown rice with intact embryos, and then the control program controls whether the flap 8 is opened or not, thereby realizing the screening of brown rice with intact embryos and brown rice with incomplete embryos.

[0099] The rice drop bin 15, the light strip 7, the flap 8 and the separation hole 16 of this embodiment are arranged at equal circumferential angle intervals along the circumferential direction, so that rice can be loaded into the rice drop trough 4 first, the image of brown rice can be photographed by the camera 6 under the illumination of the light strip 7 and the embryo integrity of the image can be identified, and then the qualified brown rice grains can be separated under the action of the flap 8, and finally the brown rice with unqualified embryo integrity can be separated. The whole process of selecting brown rice grains is smooth and efficient.

[0100] The cover body 5 used in this embodiment can prevent interference caused by external light during the shooting process, thereby ensuring the clarity of the captured image and the consistency of the shooting. In other words, the present invention uses the cover body to buckle on the frame, and during the shooting process, it plays the function of shielding the external light source, avoiding the influence of other light except the light source, which can improve the detection effect to a certain extent, and at the same time, reduce the parameter requirements for the shooting head, thereby reducing the cost.

[0101] Specific implementation method three: Combination Fig. 9 and Fig.10To illustrate this embodiment, this embodiment also includes a flap driving assembly, which includes a motor 10, a guide roller 12, a rope 13 and two bearing seats 11. The motor 10 and the two relatively arranged bearing seats 11 are installed on the rear end surface of the tommy plate 3, and the two axial ends of the guide roller 12 are rotatably installed on the two bearing seats 11, and the output shaft of the motor 10 is connected to one end of the guide roller 12, one end of the rope 13 is fixed on the guide roller 12, and the other end of the rope 13 is wound around the guide roller 12 and the rope end is fixedly connected to the back side of the flap 8.

[0102] With such arrangement, a signal is sent to the motor 10 through the control program in the chip to control the forward and reverse rotation and start and stop of the motor 10 , and the flap 8 is opened under the pulling of the rope 13 .

[0103] Specific implementation method four: Combination Fig. 9 and Fig.10 To illustrate this embodiment, this embodiment also includes a rotating shaft 14 and a spring. The flap 8 is rotatably mounted on the tormi plate 3 via the rotating shaft 14. The spring is sleeved on the rotating shaft 14 and abuts against the flap 8 and the flap mounting hole 9 on the tormi plate 3.

[0104] In this arrangement, the rotating shaft 14 mainly plays a role in the flexible rotation of the flap 8, and the spring plays a role in the reset of the flap 8. Other components and connection relationships are the same as any one of the specific embodiments 1 to 3.

[0105] When the flap 8 in the present embodiment is reset, the four sides of the flap 8 hit the rice-falling plate 2 , and the rice-falling plate 2 plays a role of limiting the reset of the flap 8 .

[0106] Specific implementation method five: Combination Figure 7 This embodiment further includes a rice drop bin 15, which is mounted on the rice drop plate 2. This arrangement facilitates the loading of rice into the rice drop trough 4. Other components and connection relationships are the same as any one of the specific embodiments 1 to 4.

[0107] Specific implementation method six: Combination Figure 7 This embodiment is described. The rice bin 15, the light strip 7, the flap 8 and the separation hole 16 of this embodiment are arranged at equal circumferential angle intervals along the circumferential direction. In this way, detection is first performed and then separation is performed. The light strip 7 is used as a backlight, so that the backlight area is photographed by a camera, and the integrity of the brown rice embryo is detected and separated through algorithm recognition. The other components and connection relationships are the same as any one of the specific embodiments one to four.

[0108] Combination Figures 6 to 10 The working principle of the seed-by-seed selection device of the present invention is described as follows:

[0109] The present invention is provided with a driving main motor in the shell 1, and the output shaft of the driving main motor passes through the center of the rice-dropping plate 3 and is fixedly connected with the center hole of the rice-dropping plate 2. The driving main motor is started to drive the rice-dropping plate 2 to rotate. During the rotation of the rice-dropping plate 2, the brown rice in the rice-dropping bin 15 enters the rice-dropping trough 4, and the rice-dropping plate 2 continues to rotate. When the rice-dropping trough 4 rotates to the light belt 7, the camera 6 takes a picture under the light transmission of the light belt 7, and identifies the integrity of the embryo through the chip, and sends instructions to the motor 10 under the control of the control program. The rotation of the motor 10 controls whether the flap 8 is opened or not. At this time, the brown rice with intact embryos after identification is separated and received by the first rice receiving box located directly below the flap 8. Fig.11 It can be seen that the photo shows a germination test of brown rice with high embryo integrity after grain-by-grain screening. Due to the screening, the average sprout length reaches 40mm-50mm.

[0110] If the brown rice is found to be unqualified after identification, the flap 8 is closed and remains stationary, and the unqualified brown rice continues to move forward and falls at the separation hole 16, and is received by the second rice receiving box located directly below the separation hole 16, thereby achieving separation and screening of qualified and unqualified brown rice grains. Fig.12 From the photo, we can see that the average sprout length of the unqualified brown rice screened out is only 15-20mm, which is less than half of the sprout length of qualified brown rice.

[0111] The above embodiments are only used to illustrate the technical solutions of the present invention, rather than to limit the same. Although the present invention has been described in detail with reference to the aforementioned embodiments, those skilled in the art should understand that the technical solutions described in the aforementioned embodiments may still be modified, or some of the technical features may be replaced by equivalents. However, these modifications or replacements do not deviate the essence of the corresponding technical solutions from the spirit and scope of the technical solutions of the embodiments of the present invention.

Claims

1. A method for detecting the integrity of embryos of brown rice before it is used as rice seeds, characterized in that: It includes the following steps: S1, obtaining a color image of brown rice grains when a light source is transmitted through the brown rice grains; S2, preprocessing the color brown rice grain image to obtain a preprocessed color image; S3, converting the preprocessed color image into a grayscale image; S4, performing light correction on the grayscale image; S5, based on the grayscale image after light correction, based on the first threshold, utilize threshold segmentation method to segment the non-brown rice grain area and the brown rice grain area of ​​each brown rice; For each brown rice grain area, subsequent processing is performed on the brown rice grain area one by one; S6. For the brown rice grain region in the grayscale image, based on the second threshold, using a threshold segmentation method to segment the embryo region in the brown rice grain region, and obtaining an outline of the embryo region, which is recorded as a first outline; S7, mapping the brown rice grain area in the grayscale image back to the color image of step S2, and extracting the brown rice grain area in the filtered color image; Then, the color image of the brown rice grain area is converted into the HSV space, and the V component image is extracted; for the V component image, the embryo region of the brown rice grain area in the V component image is segmented by setting a third threshold and using a threshold segmentation method, and the contour of the embryo region is obtained, which is recorded as the second contour; S8, mapping the second contour to the brown rice grain area corresponding to step S6, thereby obtaining the mapped contour of the second contour and the first contour on the image, and extracting contour points based on the mapped contours of the first contour and the second contour: extracting corresponding points from the two contours, and then interpolating these points to generate new contour points, thereby obtaining a more accurate fused contour; S9, based on the fused contour in step S8, convex hull detection is performed, and if the convex hull detection passes, it is determined as the embryo area for initial inspection, otherwise it is considered that the embryo is incomplete; S10. Determine the integrity of the embryo in the initial embryo inspection area.

2. The method for detecting the embryo integrity of brown rice before it is used as rice seeds according to claim 1, characterized in that: The process of preprocessing the color brown rice grain image in step S2 includes Gaussian filtering.

3. The method for detecting the embryo integrity of brown rice before it is used as rice seeds according to claim 1 or 2, characterized in that: In the process of judging the integrity of the embryo in the initial inspection embryo area in step S10, the integrity of the embryo is judged based on the area characteristics of the initial inspection embryo area.

4. The method for detecting the embryo integrity of brown rice before it is used as rice seeds according to claim 3, characterized in that: In the process of judging the integrity of the embryo based on the area characteristics of the initial inspection embryo region, when the area of ​​the embryo region is greater than the set area threshold, the embryo is judged to be complete, otherwise it is considered that the embryo is incomplete.

5. The method for detecting the embryo integrity of brown rice before it is used as rice seeds according to claim 3, characterized in that: The process of judging the integrity of the embryo based on the area characteristics of the initial embryo inspection area also includes the following steps: The brown rice grain region image corresponding to the fusion contour determined in step S8 is extracted as a whole, and stretched and transformed to a specified size; then, the area of ​​the embryo region is calculated based on the fusion contour in the stretched transformed image within the specified size range. If it is greater than the set area threshold, the embryo is judged to be complete, otherwise, the embryo is considered to be incomplete.

6. The method for detecting the embryo integrity of brown rice before it is used as rice seeds according to claim 3, characterized in that: The process of judging the integrity of the embryo based on the area characteristics of the initial embryo inspection area also includes the following steps: For the initial inspection embryo region, the area of ​​the embryo region is calculated; at the same time, the area of ​​the brown rice grain region in step S5 is calculated. If the ratio of the area of ​​the embryo region to the area of ​​the brown rice grain region is greater than the ratio threshold, the embryo is judged to be complete, otherwise it is considered that the embryo is incomplete.

7. A seed selection device for evaluating the integrity of pre-embryos of brown rice, comprising a housing (1), a rice dropping bin (15), a rice dropping plate (2) and a rice-dropping plate (3), wherein the rice-dropping plate (3) is mounted on the housing (1), the rice-dropping plate (2) is rotatably mounted on the rice-dropping plate (3), the rice-dropping plate (2) is provided with a plurality of rice-dropping grooves (4) in a circular array in a circumferential direction, and the rice-dropping bin (15) is mounted on the rice-dropping plate (2); Features: It also includes a cover body (5), a camera (6), a light strip (7) and a flap (8), wherein the light strip (7) is embedded under the towing board (3), the flap (8) can be opened and installed on the towing board (3), and a flap installation hole (9) is provided on the towing board (3) corresponding to the flap (8), and a separation hole (16) is also provided on the towing board (3), the camera (6) is installed above the rice falling board (2) through the cover body (5), and the camera (6) is located above the light strip (7), wherein a camera module is integrated in the camera (6), and the camera module is used to run the method for detecting the integrity of the embryo before the brown rice is used as rice seeds according to any one of claims 1 to 6.

8. The device for selecting the integrity of embryos before brown rice is used as rice seeds according to claim 7, characterized in that: It also includes a flap driving assembly, which includes a motor (10), a guide roller (12), a rope (13) and two bearing seats (11). A motor (10) and two bearing seats (11) arranged opposite to each other are mounted on the rear end surface of the tommy plate (3); two shaft ends of a guide roller (12) are rotatably mounted on the two bearing seats (11); an output shaft of the motor (10) is connected to one end of the guide roller (12); one end of a rope (13) is fixed to the guide roller (12); the other end of the rope (13) is wound around the guide roller (12) and the rope end is fixedly connected to the back side of the flap (8).

9. The device for selecting the integrity of embryos before brown rice is used as rice seeds according to claim 8, characterized in that: It also includes a rotating shaft (14) and a spring. The flap (8) is rotatably mounted on the tommy plate (3) via the rotating shaft (14). The spring is sleeved on the rotating shaft (14) and abuts against the flap (8) and the flap mounting hole (9) on the tommy plate (3).

10. The device for selecting the integrity of embryos before brown rice is used as rice seeds according to claim 9, characterized in that: The rice falling bin (15), the light strip (7), the flap (8) and the separation hole (16) are arranged at equal circumferential angle intervals along the circumferential direction.

Images