CN220491342U - In-situ image acquisition device for grading small red bean seeds - Google Patents

Abstract

The utility model discloses an in-situ image acquisition device for classifying small red bean seeds, which comprises an image acquisition box and a data computer, wherein positioning paper is arranged in the image acquisition box, a calibration point with primary colors and calibration specifications is arranged on the positioning paper, a shooting device is arranged opposite to the positioning paper, the shooting device is connected with the data computer, the data computer comprises a preprocessing module and a seed segmentation module, and the seed segmentation module comprises a distance map reconstruction module. The actual size and the skin color of the seeds are restored by comparing the primary colors and the calibration points of the calibration specification; the preprocessing module reduces the noise of the picture and improves the image processing precision and the picture quality; the distance map reconstruction module reduces the algorithm defects in the segmentation algorithm, eliminates the over-segmentation, and can reduce the possibility of false boundary generation of images.

Description

In-situ image acquisition device for grading small red bean seeds

Technical Field

The utility model relates to the technical field of image acquisition devices, in particular to an in-situ image acquisition device for grading small red bean seeds.

Background

The red bean is a leguminous herb plant, has higher nutritive value, has certain morphological differences such as volume, skin color and the like, is often selected and classified according to the external characteristics, but is not provided with a special pattern acquisition device for classifying the seeds of the red bean in the market, is often classified by manual labor with lower efficiency, measures the length and width data of the seeds by calipers, measures the electronic weighing, subjectively discriminates and judges according to the granularity and the personal experience knowledge of national standards, is round, cannot accurately measure the circumference and the area of the seeds by manual labor, has larger acquisition error of phenotype information of the seeds, has lower working efficiency, can easily generate eye fatigue due to the fact that the judgment of the skin color of the seeds is likely to deviate under long-time working, has high labor intensity and lower working efficiency and is easy to cause irreversible damage to the seeds in the manual picking process.

Disclosure of Invention

The utility model aims to solve the technical problems that no special pattern acquisition device for classifying the seeds of the small red beans exists in the market, the seeds are often classified by manual work with lower efficiency, but the small red beans are round, the circumference and the area of the seeds cannot be accurately measured manually, the acquisition error of the phenotype information of the seeds is larger, the working efficiency is lower, and the in-situ image acquisition device for classifying the seeds of the small red beans is provided for overcoming the defects in the prior art.

In order to achieve the purpose, the utility model adopts the following technical scheme:

an in-situ image acquisition device for grading small red bean seeds comprises an image acquisition box and a data computer;

the image acquisition box is internally fixed with positioning paper, the positioning paper is provided with a plurality of calibration points with calibration dimension specifications, wherein the calibration points can be used for comparing the actual dimension of the grain, the colors of the calibration points are primary colors, the actual surface colors of the grain can be corrected and restored according to the color difference values before and after the imaging of the calibration points, a camera is fixed on the opposite side of the positioning paper, and the camera is connected with a data computer;

the data computer comprises an image processing system, the image processing system comprises a preprocessing module and a grain segmentation module, and the grain segmentation module comprises a distance graph reconstruction module capable of reducing algorithm defects in the grain segmentation module.

Preferably, a lining board is arranged below the positioning paper, and light sources capable of supplementing light are arranged around the shooting device.

Preferably, the preprocessing module can improve the image quality and definition and reduce the redundant information of the image, and the preprocessing module comprises a graying module, a filtering module, a binarization module and a morphological operation module.

Preferably, the grain segmentation module comprises an image segmentation module, the image segmentation module comprises a grain number judgment module, a boundary strengthening module for eliminating the possibility of generating an image error boundary and a boundary generation module, and the distance map reconstruction module belongs to the image segmentation module.

Preferably, the data computer comprises a seed feature extraction system capable of extracting data such as seed shape features and color features and used for quantifying seed quality, and the seed feature extraction system comprises a feature counting module and a feature analysis module.

Preferably, the data computer comprises a seed grading system capable of screening seed quality and realizing seed grading.

Preferably, the data computer includes a data storage module capable of storing data.

The utility model has the beneficial effects that:

the utility model has simple structure, reasonable design and calibration specification, the specific real size is used as the calibration specification, the real size of the seeds can be obtained by comparing the specific real size with the size of the seeds, and the color correction can be carried out on the primary colors through chromatic aberration to restore the real surface color of the seeds; the preprocessing module can reduce the noise of pictures, improve the image processing precision and the photo quality, bring greater convenience to the subsequent image segmentation, reduce the calculation amount of a data computer and reduce the redundant information on the images; the distance map reconstruction module reduces the defects of a watershed segmentation algorithm in the segmentation algorithm, eliminates excessive segmentation, and the boundary strengthening module and the boundary generating module, can reduce the possibility of generating wrong boundaries of images, better identifies and acquires seed phenotype information, realizes more excellent image segmentation effect, and integrally improves the quality and definition of images; the seed grading system realizes digital quantization grading, greatly reduces labor intensity, greatly reduces measurement errors, improves scientificity and accuracy of measurement data, avoids eyestrain for manpower, improves accuracy and uniqueness of a seed grading result, greatly improves working efficiency of seed grading, and reduces working difficulty.

Drawings

For ease of illustration, the utility model is described in detail by the following detailed description and the accompanying drawings.

Fig. 1 is a schematic structural view of the present embodiment;

FIG. 2 is a schematic diagram of the internal algorithm flow structure of the present embodiment;

fig. 3 is a schematic diagram of a range profile reconstruction data reference simulation in the present embodiment.

In the figure: 1-an image acquisition box; 11-positioning paper; 111-calibration points; 12-a camera; 13-lining board; 14-a light source;

2-a data computer; a 21-image processing system; 211-a preprocessing module; 212-a kernel segmentation module; 2121-an image segmentation module; 22-a kernel feature extraction system; a 23-grain classification system; 24-a data saving module.

Detailed Description

The following are specific embodiments of the present utility model and the technical solutions of the present utility model will be further described with reference to the accompanying drawings, but the present utility model is not limited to these embodiments; in the following description, specific details such as specific configurations and components are provided merely to facilitate a thorough understanding of embodiments of the utility model. It will therefore be apparent to those skilled in the art that various changes and modifications can be made to the embodiments described herein without departing from the scope and spirit of the utility model. In addition, descriptions of well-known functions and constructions are omitted for clarity and conciseness.

In addition, the embodiments of the present utility model and the features of the embodiments may be combined with each other without collision.

As shown in fig. 1-3, the in-situ image collecting device for grading small red bean seeds provided by the embodiment comprises an image collecting box 1 and a data computer 2, wherein a positioning paper 11 and a camera 12 are fixed in the image collecting box 1, the camera 12 is connected with the data computer 2, the positioning paper 11 is fixed opposite to the camera 12, the positioning paper 11 is fixed at the bottom of the image collecting box 1, the camera 12 is fixed at the top of the image collecting box 1, the positioning paper 11 is about A4 size, three standard points 111 with the specification of 5mm are arranged at four corners of the positioning paper 11, the colors of the standard points 111 are respectively three primary colors of positive red, positive green and positive blue, and quantitative small red beans are placed on the positioning paper 11, when the image on the positioning paper 11 is shot by the shooting device 12 and uploaded to the data computer 2, as the phenotype information of the red bean is required to be acquired and measured, the real size of the red bean can be obtained by comparison with the real size of the standard point 111 with 5mm x 5mm standard size specification, the skin color difference between the image imaged by the shooting device 12 and the red bean can be calculated by the standard point 111 of the three primary colors, the two values are differenced by utilizing the color principle of the primary colors through the color characteristic value of the standard point 111, the color difference between the real color and the processing value is obtained, the color characteristic of the image is corrected through the color difference, and then the real skin color characteristic value is obtained;

specifically, the camera 12 may be a special industrial camera, the camera 12 adopted by the device is a three-channel industrial camera with a Maidesic MV-GE630C-T three-way, the visual field range of the camera is 224.419mm x 150mm, and the skin color of the red bean is calculated more truly and accurately by matching with the standard point 111 of the primary color, because the growth condition of crops can be calculated by means of the surface characteristics of the crops, the red bean is also calculated, the phenotype information of grains is obtained according to the volume size and the skin color of the red bean, and the grains of the red bean can be classified according to the phenotype information, for example, the red bean with different grades such as excellent, good, qualified and unqualified red bean can be selected according to the individual use requirements;

specifically, the whole image acquisition box 1 is black, the material of choice is the light-tight panel, avoid the external interference to image acquisition, for example colored light source or unknown particulate matter, influence the computer and to the collection measurement and the calculation of seed phenotype size, the below of location paper 11 is equipped with welt 13, welt 13 can assist the location paper 11 to fix in the inside of image acquisition box 1, welt 13 includes glass board and back light plate, the glass board is fixed in the top of back light plate, when the camera 12 shoots the red bean, glass board and back light plate can cooperate the camera 12 to show the image of shooting more clear bright, the outside annular light source 14 that is equipped with of camera 12 gives the inside light source 14 of picture acquisition box, the colour temperature of annular light source 14 is adjustable, camera 12 and annular light source 14 can be replaced with a smart mobile phone under the limited condition, the shooting position of gathering once, aperture luminance, focal length etc. need be fixed, in order to guarantee the fixed of image acquisition environment.

Further, the picture shot by the shooting device 12 is transmitted to the data computer 2, the original image is preprocessed, the data computer 2 comprises an image processing system 21, a grain feature extraction system 22 and a grain grading system 23, the image processing system 21 comprises a preprocessing module 211, the preprocessing module 211 can improve the quality and definition of the image and reduce the redundant information of the image, and the preprocessing module 211 comprises a block graying module, a filtering module, a binarization module and a morphological operation module;

specifically, the graying module can convert a shot color image into a gray image, so as to simplify a pixel matrix and improve operation speed, because the pixel point change range of the color image is far greater than that of the gray image for a computer;

specifically, the filtering module can inhibit image noise under the condition of retaining detailed features of grains as much as possible, which is also an indispensable operation of the preprocessing module 211, and the image filtering module can emphasize some features in some images or remove some unnecessary parts in the images, and can also well protect the shape and size of grains in the images and the specific geometric structure features to be detected, and the effectiveness and reliability of subsequent image analysis can be directly affected by the quality of the processing result of the image filtering module;

specifically, the binarization module can simplify the processing of the image, the whole image shows obvious black and white effect, the data volume in the image is greatly reduced, the outline of the red bean seeds is highlighted, the contrast is stronger, the image is clearer, in the binarization process of the image, the pixel point with the gray value lower than the threshold value is converted into 0, the pixel point higher than the threshold value is converted into 1, the noise point in the image is removed, and the image quality is improved;

specifically, the morphological operation module can quantitatively describe the shape and structure of the grain in the image, mainly extracts the image component of the grain phenotype drawing area from the image, so that the shape and characteristics of the grain can be grasped by subsequent recognition work.

Further, the image after preprocessing needs to be subjected to image segmentation, namely, the image is divided into a plurality of specific areas with unique properties and specific targets are proposed, the image processing system 21 comprises a grain segmentation module 212, the grain segmentation module 212 comprises an image segmentation module 2121 and a segmentation algorithm, the segmentation algorithm is an improved watershed segmentation algorithm applicable to red beans, and the segmentation algorithm can be used for identifying specific phenotype information to be detected of grains in the image;

specifically, the image segmentation module 2121 comprises a grain number judgment module, a distance map reconstruction module, a boundary strengthening module and a boundary generation module, the grain number judgment module can judge the grain number of red beans in an image, the watershed segmentation algorithm itself is to conduct distance map transformation on an original image to generate a distance map, the boundary is directly generated on the basis of the generated distance map, but the segmentation error is easy to be caused due to the hard defect of the watershed segmentation algorithm, the root cause of the segmentation error is the uneven gray level of an image pixel, more pseudo extremum is generated when the watershed segmentation algorithm is conducted, the pseudo extremum can be eliminated through the distance map reconstruction module, the phenomenon of excessive segmentation is further reduced, a better image segmentation effect is realized, as shown in fig. 3, for red beans 2, two points a and b are two gray level minima in the boundary, the false minimum b can be generated at the position c by directly conducting the watershed segmentation algorithm segmentation, the purpose is to eliminate the false minimum b, the mask image m which is completely the same as the hard defect of the watershed segmentation algorithm, and for the e point corresponding to be processed b point, then the false gray level value is only can be generated at the position of the boundary, and the two gray level values of the two points mu are placed in the boundary 2, and the two extreme values mu are small-level curve can be generated at the boundary;

specifically, because the watershed segmentation algorithm is sensitive to weak edges, an error boundary can be generated due to weak change of gray level, the boundary strengthening module is used for introducing a boundary strengthening algorithm to identify the boundary, and overlapping the boundary generated by the segmentation algorithm to realize image boundary strengthening, and the boundary generating module is mainly used for weakening the error boundary to generate a correct boundary, so that the effect of image segmentation is integrally improved, the hard defect of the watershed segmentation algorithm is eliminated, and the possibility of generating the image error boundary is eliminated.

Further, the image is required to be subjected to feature extraction after the image segmentation is completed, the converted seed phenotype information data is subjected to feature extraction, the extracted data features are used for quantifying the seed quality, and the seed feature extraction system 22 comprises a feature counting module and a feature analysis module;

specifically, the characteristic counting module is mainly used for counting small red bean seeds, comparing and recording the real morphological characteristics and color characteristics of the seeds, completing the data analysis of the seed phenotype information by the recorded data through the characteristic analysis module,

specifically, the feature analysis module is mainly used for carrying out statistical analysis on extracted data features, automatically carrying out parameter calculation on morphological features such as length, width, circumference and the like of each obtained seed, carrying out statistical analysis on a plurality of mathematical features such as mean value, variance and the like of seed feature values in an image, and calculating statistical parameters such as variance, mean value and the like on color parameters of the extracted single-seed red beans, wherein the obtained data features are mainly used for the next step of grading.

Further, the grain grading system 23 mainly performs dimension reduction on the extracted numerous characteristics, screens out main characteristics for determining grain quality, calculates quality calculation values among different grains according to weight relation data set in the data computer 2, compares collected phenotype data with grading judgment data for the small red bean grains in the data computer 2 in advance, grades all small red beans in batches, compares measured accurate data with judgment data, completes quality judgment on the grains with different qualities, realizes grading, compares manual judgment and grading, removes errors of manual measurement, accelerates accuracy and working efficiency of measurement judgment grading, reduces a great deal of manual labor, simultaneously avoids damage to grain epidermis and eyestrain of workers in the small red bean picking process, and reduces working intensity and working difficulty of work.

Further, the data computer 2 includes a data storage module 24, where the data storage module 24 stores all the detected and calculated data in the data computer 2, and stores the data such as the original image, the preprocessed image, the image after image segmentation, the feature statistics, the classification result, and the like in a local folder, so as to store the data.

The operation flow of the utility model is as follows:

the first step: the device is started, the camera 12 and the light source 14 are connected, the data computer 2 is connected, and shooting data of the camera 12, such as focal length, aperture and the like, are primarily adjusted;

and a second step of: tiling quantitative small red beans on the positioning paper 11, adjusting the camera 12 until the boundary of the small red beans can be clearly shot, stopping adjusting, operating the camera 12 to acquire images and recording shooting parameters of the camera 12, and ensuring that the shooting parameters at the same time are the same;

third section: the collected images are transmitted to a data computer 2 for preprocessing, and sequentially pass through an image graying module, an image filtering module, an image binarization module and an image morphological operation module;

fourth step: image segmentation is carried out on the preprocessed image, and the image segmentation is realized by matching with a segmentation algorithm through a grain number judging module, a distance graph reconstruction module, a boundary strengthening module and a boundary generating module;

fifth step: the image after image segmentation is subjected to feature extraction aiming at the grain phenotype information, the grain phenotype data information extraction and analysis are completed through a feature counting module and a feature analysis module, and each grain is subjected to judgment and grading through a grain grading system 23;

sixth step: the collected data such as measurement data, calculation results, grading results and the like are all stored in a local folder of the data computer 2;

seventh step: collecting the next batch of red bean images, and repeating the operations from the second step to the sixth step;

eighth step: the camera 12 is turned off, the light source 14 is turned off, the connection with the data computer 2 is disconnected, and the computer is turned off.

It is noted that the terminology used herein is for the purpose of describing particular embodiments only and is not intended to be limiting of example embodiments in accordance with the present application. As used herein, the singular is also intended to include the plural unless the context clearly indicates otherwise, and furthermore, it is to be understood that the terms "comprises" and/or "comprising" when used in this specification are taken to specify the presence of stated features, steps, operations, devices, components, and/or combinations thereof.

In the description of the present application, it should be understood that the azimuth or positional relationship indicated by the terms "upper", "lower", etc. are based on the azimuth or positional relationship shown in the drawings, and are merely for convenience of description of the present application and to simplify the description, and do not indicate or imply that the apparatus or element referred to must have a specific azimuth, be configured and operated in a specific azimuth, and thus should not be construed as limiting the present application.

Furthermore, the terms "first," "second," and the like, are used for descriptive purposes only and are not to be construed as indicating or implying a relative importance or implicitly indicating the number of technical features indicated. Thus, a feature defining "a first" or "a second" may explicitly or implicitly include at least one such feature. In the description of the present application, the meaning of "plurality" is at least two, such as two, three, etc., unless explicitly defined otherwise.

The specific embodiments described herein are offered by way of example only to illustrate the spirit of the present application. Those skilled in the art may make various modifications or additions to the described embodiments or substitutions in a similar manner without departing from the spirit of the utility model or exceeding the scope of the utility model as defined in the accompanying claims.

Claims (7)

1. An in-situ image acquisition device for grading small red bean seeds, which is characterized by comprising: an image acquisition box (1) and a data computer (2);

the device is characterized in that positioning paper (11) is fixed in the image acquisition box (1), a plurality of calibration points (111) with calibration dimensions capable of comparing the actual dimensions of the seeds are arranged on the positioning paper (11), the colors of the calibration points (111) are primary colors, the actual skin colors of the seeds can be corrected and restored according to the color difference values before and after the imaging of the calibration points (111), a camera (12) is fixed on the opposite side of the positioning paper (11), and the camera (12) is connected with a data computer (2);

the data computer (2) comprises an image processing system (21), the image processing system (21) comprises a preprocessing module (211) and a grain segmentation module (212), and the grain segmentation module (212) comprises a distance graph reconstruction module capable of reducing algorithm defects in the grain segmentation module (212).

2. An in-situ image acquisition device for classifying small red bean seeds according to claim 1, wherein a lining plate (13) is arranged below the positioning paper (11), and light sources (14) capable of supplementing light are arranged around the camera (12).

3. An in-situ image acquisition device for grading small red bean seeds according to claim 2, wherein the preprocessing module (211) can improve image quality and definition and reduce redundant information of images, and comprises a graying module, a filtering module, a binarizing module and a morphological operation module.

4. An in-situ image acquisition device for red bean grain classification as claimed in claim 3, wherein the grain segmentation module (212) comprises an image segmentation module (2121), the image segmentation module (2121) comprises a grain number judgment module, a boundary strengthening module for eliminating the possibility of generating an image error boundary and a boundary generation module, and the distance map reconstruction module belongs to the image segmentation module (2121).

5. An in-situ image acquisition device for red bean kernel classification as in claim 4 wherein said data computer (2) comprises a kernel feature extraction system (22) capable of extracting kernel shape feature, color feature, etc. data and for quantifying kernel quality, said kernel feature extraction system (22) comprising a feature counting module, a feature analysis module.

6. An in situ image acquisition device for red bean kernel classification as in claim 5 wherein said data computer (2) includes a kernel classification system (23) capable of screening kernel quality to effect kernel classification.

7. An in situ image acquisition device for red bean kernel classification as in claim 6, wherein the data computer (2) comprises a data storage module (24) capable of data storage.