Brown rice full-surface defect real-time identification system

Abstract

The invention discloses a brown rice full-surface defect real-time identification system, which comprises a machine base, a camera, a light source system, a sample carrying table and a sample carrier, and is characterized in that the light source system comprises an annular light source; the camera is connected with the image processing system; the sample carrier is provided with an accommodating groove for accommodating brown rice; the sample carrying table and the sample carrier can transmit light; the light source system further comprises an area light source arranged below the sample carrying table; in the overlook direction, the groove bottom of the containing groove is in a rectangular shape, and the two long edges of the groove bottom are respectively provided with an inclined face extending upwards and outwards. A reflecting layer is arranged on the inclined surface. According to the invention, the arrangement mode of the upper light source and the lower light source is adopted, so that the brown rice in the accommodating groove can be fully irradiated, and the surface and internal defects of the brown rice can be fully shown; according to the special structural design of the containing groove, it can be guaranteed that light can penetrate into the containing groove, meanwhile, the camera can collect images of the two sides of the grain in the containing groove through the reflecting layers on the two slopes, and all-directional collection of the brown rice image is achieved.

Description

Technical Field

[0001] The present invention relates to the technical field of image acquisition and defect recognition of grain defects, and in particular to a real-time recognition system for defects on the entire surface of brown rice. Background Art

[0002] At present, the selection and purchase of rice seeds mainly relies on the evaluation of appearance quality indicators. Appearance quality indicators can help consumers judge the appearance quality and taste of rice. The mechanical force generated by special equipment (rice mill) can be used to remove the rice husks. The common appearance indicators after milling are mainly the following: unmilled rice grains, broken rice grains, cracked rice grains, insect-eaten rice grains, moldy rice grains, normal rice grains, etc. By judging the surface quality of brown rice after milling, high-quality rice can be accurately distinguished.

[0003] Traditional methods for evaluating rice surface quality indicators are time-consuming, subjective, and inconsistent. Machine vision-based recognition systems mostly focus on identifying single-sided surfaces, single seeds, or single appearance quality indicators. There is little research on batch detection, full-surface detection, and detection of multiple appearance quality indicators.

[0004] In the prior art, patent CN105430350A provides a grain grain image acquisition system, which scans the upper and lower surfaces of grain and performs grain detection based on the images. This method uses scanning means to capture images and requires two upper and lower scanners, which is costly.

[0005] Patent CN108896569A discloses a device for detecting full-surface defects on rolling elements. This device, primarily used for inspecting cylindrical objects, employs multiple through-grooves provided on a running belt. The rolling elements are placed within these grooves. As the belt rotates, friction between the rolling elements within the grooves and the fixed plates at the bottom of the grooves causes them to roll. This allows a camera to capture images of different surfaces of the rolling elements. This device is complex and costly. Furthermore, since unpolished rice is not a standard rolling element, it may not produce the expected rolling motion. Furthermore, the camera must capture multiple sets of images for the same batch of rolling elements, requiring a large amount of data and increasing computing power requirements. Summary of the Invention

[0006] Purpose of the invention: In order to overcome the deficiencies in the prior art, the present invention provides a real-time brown rice surface defect recognition system with a simple structure, low cost, and the ability to comprehensively capture images of the brown rice surface for defect recognition.

[0007] Technical solution: To achieve the above-mentioned purpose, the present invention provides a real-time identification system for defects on the entire surface of brown rice, which includes a base, a camera, a light source system, a sample carrier, and a sample carrier, wherein the camera is placed above the sample carrier, the light source system includes an annular light source arranged coaxially with the lens of the camera; the camera is connected to an image processing system; and the sample carrier has a accommodating tank for accommodating brown rice.

[0008] The sample carrier and the sample carrier are both light-transmissive, and the light source system further comprises a surface light source disposed below the sample carrier;

[0009] The bottom of the receiving groove is rectangular when viewed from above, and its two long sides each have an upwardly extending, outwardly extending inclined surface. A reflective layer is applied to each inclined surface, forming a mirrored surface capable of reflecting light. The walls extending upward from the two short sides of the bottom and the bottom of the receiving groove are both light-transmissive.

[0010] Furthermore, the inclination angle of the inclined surface relative to the plane is not less than 45°.

[0011] Furthermore, the sample carrier includes a base made of acrylic material and a trapezoidal member made of glass. The cross-section of the trapezoidal member is an isosceles trapezoid, and the two waist surfaces of the trapezoidal member are covered with the reflective layer. The base has a groove corresponding to the location of each of the accommodating grooves, and also has a trapezoidal groove for the trapezoidal member to be inserted into. The groove is connected to the trapezoidal groove. After the trapezoidal member is inserted into the trapezoidal groove, the local waist surfaces of the two trapezoidal members on both sides of each groove are formed. In this structure, the base is easy to process and ensures light guidance. The design of the trapezoidal member ensures the flatness and deformation resistance of the reflective layer.

[0012] Furthermore, the base includes two parallel columns; the camera, sample stage, ring light source, and surface light source are each adjustable in height along the columns, and their horizontal positions are each adjustable perpendicular to the columns. This facilitates the relative positioning of the components, allowing for fine-tuning of the light source illumination and the distance between the camera and the sample, achieving optimal image acquisition, reducing the recognition difficulty of the image processing system, and improving recognition accuracy.

[0013] Furthermore, the sample carrier is connected to the column through an adjustment mechanism; the adjustment mechanism includes a connecting block that is threaded on the column, and also includes a flat column that can adjust its position horizontally relative to the connecting block and a U-shaped frame connecting the flat column, and the sample carrier is installed in the U-shaped frame; the connecting block can be fixed relative to the column under the action of a first fixing screw, and the flat column can be fixed relative to the connecting block under the action of a second fixing screw.

[0014] Furthermore, the U-shaped frame includes a horizontal bar connecting the horizontal column, and two longitudinal bars connecting the horizontal bar; the cross-section of the longitudinal bar is L-shaped, and it has a vertical plate portion and a horizontal plate portion, and the horizontal plate portion supports the edge of the sample carrier, and the horizontal plate portion has a first strip groove, and the screw connecting the sample carrier can adjust the position along the first strip groove.

[0015] Furthermore, a positioning frame is mounted above the sample carrier, and the sample carrier is removably mounted on the positioning frame. The vertical plate portion has a second strip groove, and long screws are mounted on both sides of the positioning frame. The long screws pass through the second strip grooves on the longitudinal rods on the corresponding sides, and each long screw is screwed onto a nut for fixing the position. The longitudinal position of the positioning frame can be adjusted by adjusting the position of the long screws in the direction of the second strip grooves, and the lateral position of the positioning frame can be adjusted by adjusting the position of the nuts along the axial direction of the long screws. In this way, the position of the positioning frame can be conveniently and infinitely adjusted to position the positioning frame in the center of the camera's field of view.

[0016] Furthermore, positioning members are fixed on both sides of the sample carrier respectively; when the sample carrier is placed on the positioning frame, the two positioning members are respectively placed on both sides of the positioning frame.

[0017] Beneficial effects: The real-time system for identifying defects on the entire surface of brown rice of the present invention has the following beneficial effects:

[0018] (1) The real-time defect recognition system of the present invention adopts the layout of upper and lower light sources to fully illuminate the brown rice in the storage tank, so that the surface and internal defects of the brown rice can be fully displayed in the collected image; the special structural design of the storage tank can ensure that the light of the surface light source can penetrate the storage tank, while the camera can collect images of the two sides of the grain in the storage tank through the reflective layers on the two inclined surfaces, realizing the full-scale collection of the brown rice image, and the image processing system can perform comprehensive recognition of the defects of the brown rice based on the image.

[0019] (2) The sample carrier is manufactured by assembling a base and a trapezoidal part. Its structural design and assembly method can greatly reduce costs, ensure light transmittance, and ensure that the reflective surface is flat and not easily deformed, so that the reflective surface can fully reflect the image of the side of the brown rice to the camera.

[0020] (3) The sample carrier is positioned based on the positioning frame and the positioning parts, which can achieve rapid positioning and rapid disassembly of the sample carrier. The connection structure between the positioning frame and the U-shaped frame can facilitate fine adjustment of the position of the sample carrier in the camera's field of view to ensure the image acquisition effect. BRIEF DESCRIPTION OF THE DRAWINGS

[0021] Figure 1This is the structural diagram of the real-time recognition system for all surface defects of brown rice;

[0022] Figure 2 A structural diagram of a first combination of a sample carrier, a sample carrier, and an adjustment mechanism;

[0023] Figure 3 A second structural diagram of the sample carrier, the sample carrier, and the adjustment mechanism;

[0024] Figure 4 This is the exploded structure diagram of the sample carrier;

[0025] Figure 5 is a structural diagram of the sample carrier;

[0026] Figure 6 This is a partial diagram of the light principle of the sample carrier when acquiring images;

[0027] Figure 7 This is the installation structure diagram of the positioning model.

[0028] In the figure: 1-camera; 2-sample carrier; 3-ring light source; 4-surface light source; 5-sample carrier; 51-base; 51a-trapezoidal groove; 51b-groove; 52-trapezoidal member; 5a-accommodating groove; 5b-inclined surface; 6-base; 61-column; 7-adjustment mechanism; 71-connecting block; 72-parallel column; 731-cross bar; 732-longitudinal bar; 73a-first strip groove; 73b-second strip groove; 8-positioning frame; 81-long screw; 82-nut; 9-positioning member. DETAILED DESCRIPTION

[0029] The present invention will be further described below with reference to the accompanying drawings.

[0030] like Figure 1 The real-time recognition system for defects on the entire surface of brown rice shown in the figure comprises a base 6, a camera 1, a light source system, a sample carrier 2, and a sample carrier 5. The camera 1 is placed above the sample carrier 2, and the light source system comprises an annular light source 3 coaxially arranged with the lens of the camera 1; the camera 1 is connected to an image processing system; and the sample carrier 5 has a receiving groove 5a for receiving brown rice.

[0031] The sample carrier 2 and the sample carrier 5 are both light-transmissive, and the light source system further includes a surface light source 4 placed below the sample carrier 2;

[0032] In the top view, the bottom of the accommodating groove 5a is rectangular. Figure 5As shown, the two long sides of the groove bottom each have an upward and outward extending inclined surface 5b; the inclined surface 5b has a reflective layer, so that the two inclined surfaces 5b form a mirror surface that can reflect light. The wall surfaces extending upward from the two short sides of the groove bottom and the groove bottom of the groove receiving groove 5a are both light-transmissive.

[0033] The real-time defect recognition system of the present invention adopts the layout of upper and lower light sources to fully illuminate the brown rice in the receiving tank 5a, so that the surface and internal defects of the brown rice can be fully displayed in the collected image; the special structural design of the receiving tank 5a can ensure that the light of the surface light source 4 can penetrate the receiving tank 5a, such as Figure 6 As shown, the camera 1 can capture images of the two sides of the grain in the accommodating tank 5a through the reflective layers on the two inclined surfaces 5b, thereby achieving all-round capture of the brown rice image. The image processing system can fully identify the defects of the brown rice based on the image.

[0034] Preferably, the inclination angle of the inclined surface 5b relative to the plane is not less than 45°.

[0035] Preferably, if Figure 4 and Figure 5 As shown, the sample carrier 5 comprises a base 51 made of acrylic material and a trapezoidal member 52 made of glass. The cross-section of the trapezoidal member 52 is an isosceles trapezoid, and the two waist surfaces of the trapezoidal member 52 are covered with the reflective layer. The base 51 has a groove 51b corresponding to the location of each of the accommodating grooves 5a, and also has a trapezoidal groove 51a for the trapezoidal member 52 to be inserted into. The groove is connected to the trapezoidal groove. After the trapezoidal member 52 is inserted into the trapezoidal groove, the two local waist surfaces of the trapezoidal member 52 on either side of each groove, and the grooves in the same row share the inclined surface of the trapezoidal member 52. In this structure, the base 51 is easy to process and ensures light guidance. The design of the trapezoidal member 52 ensures the flatness and deformation resistance of the reflective layer. The sample carrier 5 is manufactured by assembling the base 51 and the trapezoidal member 52. This structural design and assembly method can significantly reduce costs, ensure light transmittance, and ensure that the reflective surface is flat and not easily deformed, so that the reflective surface can fully reflect the image of the side of the brown rice to the camera 1.

[0036] Preferably, the base 6 includes two parallel columns 61; the camera 1, sample stage 2, annular light source 3, and surface light source 4 are each adjustable in height along the columns 61 and in horizontal position perpendicular to the columns 61. This facilitates adjustment of the relative positions of the components, allowing for fine-tuning of the light source illumination and the distance between the camera 1 and the sample, achieving optimal image acquisition, reducing the recognition difficulty of the image processing system, and improving recognition accuracy.

[0037] Preferably, the sample carrier 2 is connected to the column 61 via an adjustment mechanism 7; Figure 2 As shown, the adjustment mechanism 7 includes a connecting block 71 that is threaded on the column 61, and also includes a flat column 72 that can adjust its position horizontally relative to the connecting block 71 and a U-shaped frame connecting the flat column 72, and the sample carrier 2 is installed in the U-shaped frame; the connecting block 71 can be fixed relative to the column 61 under the action of a first fixing screw 74, and the flat column 72 can be fixed relative to the connecting block 71 under the action of a second fixing screw 75.

[0038] Preferably, the U-shaped frame includes a crossbar 731 connecting the horizontal columns 72, and also includes two longitudinal bars 732 connecting the crossbar 731; the cross section of the longitudinal bar 732 is L-shaped, which has a vertical plate portion and a horizontal plate portion, such as Figure 3 As shown, the horizontal plate portion supports the edge of the sample carrier 2 and has a first strip groove 73a on the horizontal plate portion. The screw connecting the sample carrier 2 can adjust the position along the first strip groove 73a.

[0039] The camera 1 , the annular light source 3 and the surface light source 4 are also adjusted relative to the column 61 using a mechanism similar to the above-mentioned adjustment mechanism 7 , which will not be described in detail here.

[0040] Preferably, if Figure 7 As shown, a positioning frame 8 is mounted above the sample carrier 2, and the sample carrier 2 is detachably mounted on the positioning frame 8. The vertical plate portion has a second strip groove 73b, and long screws 81 are mounted on both sides of the positioning frame 8. The long screws 81 pass through the second strip groove 73b on the longitudinal rod 732 on the corresponding side, and each of the long screws 81 is screwed with a nut 82 for fixing the position. By adjusting the position of the long screw 81 in the extension direction of the second strip groove 73b, the longitudinal position of the positioning frame 8 can be adjusted, and by adjusting the position of the nut 82 along the axial direction of the long screw 81, the lateral position of the positioning frame 8 can be adjusted. In this way, the position of the positioning frame 8 can be conveniently and infinitely adjusted so that the positioning frame 8 is located in the center of the field of view of the camera 1.

[0041] Preferably, positioning members 9 are fixed on both sides of the sample carrier 5 ; when the sample carrier 5 is placed on the positioning frame 8 , two positioning members 9 are respectively placed on both sides of the positioning frame 8 .

[0042] By positioning the sample carrier 5 based on the positioning frame 8 and the positioning part 9, the sample carrier 5 can be quickly positioned and quickly disassembled and assembled, and the connection structure between the positioning frame 8 and the U-shaped frame 73 can facilitate fine adjustment of the position of the sample carrier 5 in the field of view of the camera 1 to ensure the image acquisition effect.

[0043] In the present invention, the image processing system can recognize the image captured by the camera 1 based on models such as YOLOv5, YOLOv7, and Faster R-CNN to achieve defect recognition.

[0044] The above is only a preferred embodiment of the present invention. It should be pointed out that for ordinary technicians in this technical field, several improvements and modifications can be made without departing from the principles of the present invention. These improvements and modifications should also be regarded as the scope of protection of the present invention.

Claims

1. A real-time recognition system for the entire surface defect of brown rice, comprising a base (6), a camera (1), a light source system, a sample carrier (2) and a sample carrier (5), wherein the camera (1) is placed above the sample carrier (2), the light source system comprises an annular light source (3) coaxially arranged with a lens of the camera (1); the camera (1) is connected to an image processing system; the sample carrier (5) has a receiving groove (5a) for receiving brown rice; and the system is characterized in that: The sample carrier (2) and the sample carrier (5) are both light-transmissive, and the light source system further comprises a surface light source (4) disposed below the sample carrier (2); In a top view, the bottom of the accommodating groove (5a) is rectangular, and two long sides of the bottom are respectively provided with inclined surfaces (5b) extending upward and outward; and a reflective layer is provided on the inclined surfaces (5b).

2. brown rice full surface defect real-time identification system according to claim 1, is characterized in that, The inclination angle of the inclined surface (5b) relative to the plane is not less than 45°.

3. brown rice full surface defect real-time identification system according to claim 1, is characterized in that, The sample carrier (5) comprises a base (51) made of acrylic material, and also comprises a trapezoidal piece (52) made of glass material, wherein the two waist surfaces of the trapezoidal piece (52) are covered with the reflective layer; the base (51) has a groove arranged corresponding to the position of each of the accommodating grooves (5a), and also has a trapezoidal groove for the trapezoidal piece (52) to be embedded.

4. brown rice full surface defect real-time identification system according to claim 1, is characterized in that, The base (6) includes two parallel columns (61); the camera (1), sample carrier (2), annular light source (3), and surface light source (4) can be adjusted in height along the columns (61) and can be adjusted in horizontal position in a direction perpendicular to the columns (61).

5. brown rice full surface defect real-time identification system according to claim 3, is characterized in that, The sample carrier (2) is connected to the column (61) through an adjustment mechanism (7); the adjustment mechanism (7) includes a connecting block (71) that is sleeved on the column (61), and also includes a flat column (72) that can adjust its position horizontally relative to the connecting block (71) and a U-shaped frame connected to the flat column (72), and the sample carrier (2) is installed in the U-shaped frame; the connecting block (71) can be fixed relative to the column (61) under the action of a first fixing screw (74), and the flat column (72) can be fixed relative to the connecting block (71) under the action of a second fixing screw (75).

6. brown rice full surface defect real-time identification system according to claim 4, is characterized in that, The U-shaped frame includes a cross bar (731) connected to the horizontal column (72), and also includes two longitudinal bars (732) connected to the cross bar (731); the cross section of the longitudinal bar (732) is L-shaped, and it has a vertical plate portion and a horizontal plate portion, the horizontal plate portion supports the edge of the sample carrier (2), and the horizontal plate portion has a first strip groove (73a), and the screw connecting the sample carrier (2) can adjust the position along the first strip groove (73a).

7. brown rice full surface defect real-time identification system according to claim 5, is characterized in that, A positioning frame (8) is installed above the sample carrier (2), and the sample carrier (2) is detachably installed on the positioning frame (8); a second strip groove (73b) is provided on the vertical plate portion, and long screws (81) are respectively installed on both sides of the positioning frame (8), and the long screws (81) pass through the second strip groove (73b) on the longitudinal rod (732) on the corresponding side, and each of the long screws (81) is screwed with a nut (82) for fixing the position.

8. brown rice full surface defect real-time identification system according to claim 7, is characterized in that, Positioning members (9) are fixed on both sides of the sample carrier (5); when the sample carrier (5) is placed on the positioning frame (8), the two positioning members (9) are respectively placed on both sides of the positioning frame (8).

Images