CN103234905B - A method and a device for acquiring non-redundant image information of ball-shaped fruits - Google Patents

Abstract

The invention discloses a method and a device for acquiring spherical fruit without redundant image information. Includes servo motors, 4 circumferential cameras, 1 top-view camera, trigger controller, computer, 4 cross laser markers, slot photoelectric sensors, trays and conveyor belts. The trigger controller is respectively connected with the servo motor, the slot-shaped photoelectric sensor, 4 circumferential cameras and the top-view camera. The computer is connected with 4 peripheral cameras and 1 top-view camera respectively. The fruit is loaded on trays, which are mounted on conveyor belts. The invention utilizes cross laser markings to locate boundaries between multiple fruit images, thereby improving the splicing speed and image splicing accuracy of the fruit images.

Description

Method and device for acquiring non-redundant image information of spherical fruit

technical field

The invention relates to a method and a device for acquiring image information, in particular to a method and a device for acquiring spherical fruit without redundant image information.

Background technique

With the continuous improvement of the degree of automation and the continuous development of precision agriculture, the requirements for the accuracy of fruit grading are bound to become higher and higher. Extracting an image without redundant information on the surface of the fruit, and then further judging the quality of the fruit based on the image, such as color, defects, etc., is of great significance for improving the classification accuracy of the fruit. In the research of the existing production line, the ways of collecting images of fruits through the camera mainly include:

A single camera acquires images. In the process of image acquisition, the fruit rotates itself while the production line is running on the production line, and the camera field of view includes multiple stations. Therefore, multiple images can be collected for the same fruit. The US patent "Methold and apparatus for sorting objects by color (method and device for grading objects by color) application number: 5339963" by Yang Tao et al. uses a camera to detect and classify objects on the conveyor chain. Ying Yibin and others from Zhejiang University patented the fruit quality real-time detection and grading robot system (application number: 02136377.3), the fruit real-time grading control system controlled by shift registers (application number: 02266031.3) the fruit sorting machine patent (application number : 201120140461.0), Fruit Grading Machine Based on Machine Vision (Application No.: 02295073.7), etc. describe a grading system that uses a single camera to acquire fruit surface images.

Multiple cameras acquire images simultaneously. In a multi-camera system, one method is to place the fruit in the fruit cup, triggering multiple cameras to simultaneously collect images of the fruit (Li Qingzhong, Wang Maohua. Hardware Development of Apple Automatic Grading System Based on Machine Vision. Journal of Agricultural Machinery, 2000, 31(2):56-59); one is that the fruit production line rotates itself as the production line runs, triggering the camera to collect multiple fruit images, such as the online fruit quality detection and grading system based on three camera systems designed by Zhao Jiewen, etc. Device and method (application number: 200410065216.2).

No matter which image acquisition method is used to obtain images of fruits for further detection of fruit quality, there is a problem that image information on fruit surfaces is repeatedly collected. Judging fruit quality based on images with repeated information will inevitably affect the accuracy of grading. At present, there is no method that can guarantee the acquisition of multi-surface images of spherical fruits without redundant information.

Contents of the invention

Aiming at the problem of image redundancy in the surface image acquisition process of spherical fruits, the purpose of the present invention is to provide a method and device for obtaining non-redundant image information of spherical fruits, combined with multiple cameras, cross laser emission and trigger control circuits, It is guaranteed to obtain images without redundant information on the multi-surface of spherical fruit.

The technical solution adopted by the present invention to solve its technical problems is:

1. The method for obtaining multi-surface non-redundant information images of spherical fruits:

A top-view camera and 4 peripheral cameras were used to collect 1 fruit top-view image with cross laser markings, 4 fruit circumferential images with cross laser markings, and 1 without The fruit top view image with cross laser marking and 4 fruit circumferential images without cross laser marking; 1 fruit top view image with cross laser marking and 4 fruit circumferential images with cross laser marking Correspondingly subtracted from 1 fruit top-view image without cross laser markings and 4 fruit circumferential images without cross laser markings respectively, to obtain 1 top-view laser line image and 4 circumferential laser line images; Multiply the circular area of the top-view laser line image by the fruit top-view image without cross laser markings to obtain the region of interest in the top-view image; Multiply the peripheral image of the fruit with the line to obtain the region of interest of four peripheral images; combine the region of interest of the top-view image and the region of interest of the peripheral image to obtain an image without redundant information on the fruit surface.

2. A device for acquiring multi-surface images of spherical fruits without redundant information:

The invention includes servo motors, 4 circumferential cameras, a top-view camera, a trigger controller, a computer, 4 cross laser markers, a groove-shaped photoelectric sensor, a tray and a conveyor belt.

The 4 circumferential cameras are evenly arranged in the same plane in a ring shape, the 4 cross laser markers are evenly arranged between the 4 circumferential cameras, and the transmission belt driven by the servo motor is arranged under the circumferential camera and the cross laser marker , the space angle between the transmission belt and a circumferential camera is 22.5 degrees; the servo motor drives the transmission belt to rotate, the fruit is loaded on the tray, the tray is installed on the transmission belt, and the top-view camera is installed above the fruit; the groove-shaped photoelectric sensor straddles the two sides of the transmission belt, And located in front of the forward direction of the pallet;

The trigger controller is respectively connected with the servo motor, the groove-shaped photoelectric sensor, the 4 peripheral cameras and the top-view camera, and the computer is connected with the 4 peripheral cameras and the top-view camera respectively.

The trigger controller: includes microprocessor U1, optocoupler U2, optocoupler U3, motor driver U4, optocoupler U5, relay U6, transistor Q1, diode D1, potentiometer Rp, resistor R1, resistor R2, resistor R3 , resistor R4, resistor R5 and resistor R6;

The brown lead wire and the blue lead wire of the slot-shaped photoelectric sensor are respectively connected to +24V power supply and GND1, the black lead wire of the slot-shaped photoelectric sensor is connected to the first pin of optocoupler U2 through resistor R2, the second pin of optocoupler U2 is connected to GND1, and the resistor R1 is respectively Connect to +5V power supply and pin 6 of microprocessor U1, pin 5 of optocoupler U2 connects pin 6 of microprocessor U1, pin 4 of optocoupler U2 connects to GND0, pin 19 of microprocessor U1 The pin is connected to the trigger signal input terminal of the camera, the resistor R4 is respectively connected to the +5V power supply and the first pin of the optocoupler U5, the second pin of the optocoupler U5 is connected to the 17th pin of the microprocessor U1, and the fifth pin of the optocoupler U5 is connected to +24V Power supply, resistor R5 is connected between the 4th pin of optocoupler U5 and the base of transistor Q1, the coil of relay U6 is connected between the +24V power supply and the collector of transistor Q1, the emitter of the transistor is connected to GND1, and the diode D1 is reversely connected to relay U6 At both ends of the coil, the normally open contact of the relay U6 is connected between the +3V power supply and the 4 cross laser markers, the negative poles of the power supplies of the 4 cross laser markers are connected to GND1, and the first pin of the optocoupler U3 is connected to the micro The 18th pin of the processor U1, the 2nd pin of the optocoupler U3 are connected to GND0 through the resistor R7, the 3rd pin of the optocoupler U3 is connected to the +24V power supply, and the 4th pin of the optocoupler U3 is respectively connected to the CN1 interface of the motor driver U4 2 pins and resistor R6, the 13th pin of the CN1 interface of the motor driver U4 is connected to GND1 and the resistor R6, the resistor R3 is connected between the +24V power supply and the 22nd pin of the CN1 interface of the motor driver U4, and the potentiometer Rp is connected to The 22nd pin of the CN1 interface of the motor driver U4 is connected to GND1; the TB3 interface of the motor driver U4 is connected to the servo motor.

The beneficial effects that the present invention has are:

The invention utilizes the cross laser marking to locate the boundaries between multiple fruit images, thereby improving the splicing speed and image splicing accuracy of the fruit images.

Description of drawings

Fig. 1 is a structural schematic diagram of the present invention.

Figure 2 is a top view excluding the top view camera.

Figure 3 is a schematic diagram of the trigger control circuit.

Fig. 4 is a schematic diagram of trigger control timing.

Fig. 5 is a schematic diagram of images collected by a top-view camera.

Fig. 6 is a schematic diagram of images collected by one of the peripheral cameras.

Figure 7 is the region of interest (shaded area) in the images acquired by the five cameras.

Fig. 8 is the obtained multi-surface non-redundant information image of a quasi-spherical fruit.

In the figure: 1. Servo motor, 2. 4 circumferential cameras, 3. Top-view camera, 4. Trigger controller, 5. Computer, 6. 4 cross laser markers, 7. Fruit, 8. Groove shape Photoelectric sensor, 9, pallet, 10, conveyor belt.

Detailed ways

The present invention will be further described below in conjunction with the accompanying drawings and embodiments.

As shown in Figure 1, the present invention includes servo motor 1, 4 circumferential cameras 2, top-view camera 3, trigger controller 4, computer 5, 4 cross laser markers 6, groove photoelectric sensor 8, tray 9 and conveyor belt 10 .

As shown in Figure 1 and Figure 2, the four circumferential cameras 2 are evenly arranged on the same plane in a ring shape, and the four cross laser markers 6 are evenly arranged between the four circumferential cameras 2, driven by the servo motor 1 The transmission belt 10 is arranged under the circumferential camera 2 and the cross laser marker 6, and the space angle between the transmission belt 10 and a circumferential camera 2 is 22.5 degrees; the servo motor 1 drives the transmission belt 10 to rotate, and the fruit 7 is loaded on the tray 9, and the tray 9 Installed on the transmission belt 10, the top-view camera 3 is installed above the fruit 7; the groove-shaped photoelectric sensor 8 straddles the transmission belt 10 both sides, and is positioned at the front of the pallet 9 advancing direction.

The trigger controller 4 is connected with the servo motor 1, the groove-shaped photoelectric sensor 8, the four peripheral cameras 2 and the top-view camera 3, and the computer 5 is connected with the four peripheral cameras 2 and the top-view camera 3 respectively.

As shown in Figure 3, the trigger controller 4 includes a microprocessor U1, an optocoupler U2, an optocoupler U3, a motor driver U4, an optocoupler U5, a relay U6, a transistor Q1, a diode D1, a potentiometer Rp, a resistor R1, resistor R2, resistor R3, resistor R4, resistor R5 and resistor R6;

The brown lead wire and the blue lead wire of slot-shaped photoelectric sensor 8 are respectively connected to +24V power supply and GND1, the black lead wire of slot-shaped photoelectric sensor 8 is connected to the first pin of optocoupler U2 through resistor R2, the second pin of optocoupler U2 is connected to GND1, and the resistor R1 is respectively connected to the +5V power supply and the 6th pin of the microprocessor U1, the 5th pin of the optocoupler U2 is connected to the 6th pin of the microprocessor U1, the 4th pin of the optocoupler U2 is connected to GND0, and the 4th pin of the optocoupler U2 is connected to GND0. Pin 19 is connected to the camera trigger signal input terminal, resistor R4 is respectively connected to +5V power supply and the first pin of optocoupler U5, the second pin of optocoupler U5 is connected to the 17th pin of microprocessor U1, and the fifth pin of optocoupler U5 is connected to +24V power supply, resistor R5 is connected between the 4th pin of optocoupler U5 and the base of transistor Q1, the coil of relay U6 is connected between the +24V power supply and the collector of transistor Q1, the emitter of the transistor is connected to GND1, and the diode D1 is reversely connected to The two ends of the coil of the relay U6, the normally open contact of the relay U6 are connected between the +3V power supply and the four cross laser markers (6), the negative poles of the power supply of the four cross laser markers 6 are connected to GND1, and the optocoupler U3 The first pin is connected to the 18th pin of the microprocessor U1, the second pin of the optocoupler U3 is connected to GND0 through the resistor R7, the third pin of the optocoupler U3 is connected to the +24V power supply, and the fourth pin of the optocoupler U3 is respectively connected to the motor driver The second pin of the CN1 interface of U4 and resistor R6, the 13th pin of the CN1 interface of the motor driver U4 is connected to GND1 and the resistor R6, and the resistor R3 is connected between the +24V power supply and the 22nd pin of the CN1 interface of the motor driver U4 Between, the potentiometer Rp is connected to the 22nd pin of the CN1 interface of the motor driver U4 and GND1; the TB3 interface of the motor driver U4 is connected to the servo motor 1.

As shown in Figure 1, during work, fruit 7 is loaded on the tray 9. The image of fruit 7 was acquired by double exposure method. Driven by the servo motor 1, the transmission belt 10 drives the tray 9 to move from left to right.

(1) As shown in Figure 4, when the fruit 7 reaches the groove-shaped photoelectric sensor 8, the groove-shaped photoelectric sensor 8 generates a high level, which makes the fifth pin of the optocoupler U2 change to a low level, triggering an interrupt of the microprocessor U1.

(2) As shown in Figure 4, the microprocessor U1 interrupt service processing program first turns off the interrupt, so that the 18th pin of the microprocessor U1 becomes low level, and the servo motor 1 stops moving; after a delay of T0, the microprocessor The 17th pin of U1 becomes low level, which triggers the cross laser marker 6 to emit laser; after a delay of T1, the 19th pin of microprocessor U1 outputs high level, which triggers the acquisition of the top-view camera 3 and the circumferential camera 2 Fruit 7's 1 fruit top view image with cross laser marking and 4 fruit circumferential images with cross laser marking.

(3) Trigger the top-view camera 3 and the peripheral camera 2 to send the photos to the computer 5 .

(4) After the microprocessor U1 delays T2, the 17th pin of the microprocessor U1 outputs a high level, and the cross laser marker 6 is turned off; at the same time, the 19th pin of the microprocessor U1 outputs a low level; delay again After time T3, the 19th pin of the microprocessor U1 outputs a high level, triggering the top-view camera 3 and the circumferential camera 2 to collect 1 fruit top-view image without cross laser marking and 4 fruit top-view images without cross Circumferential image of fruit with laser marking.

(5) Trigger the top-view camera 3 and the peripheral camera 2 to send the photos to the computer 5 .

(6) After another delay of T3, the 19th pin of the microprocessor U1 becomes low level, and at the same time, the 18th pin of the microprocessor U1 outputs a high level, so that the servo motor 1 starts to move;

As shown in Figure 5, the top-view images of fruit 7 with cross laser markings (shown in Figure 5 (a)) and fruit top-view images without cross laser markings ( Figure 5 (b) shown). The boundary of the inner circle is the line formed by the lasers emitted by the 4 cross laser markers (shown by the dotted circle in Figure (a)), and the outer circle is the boundary of Fruit 7 (shown by the solid line circle in Figure (a)) . The intersecting vertical lines (shown by dotted lines in Figure (a)) are formed by the lasers emitted by the four cross laser markers. Subtract the fruit top view image with cross laser marking (shown in Figure 5 (a)) and the fruit top view image without cross laser marking (shown in Figure 5 (b)), and extract the top View the laser line image. Multiply the extracted circular area of the top-view laser line by the top-view image of the fruit without cross laser markings (shown in (b) in Figure 5) to obtain the region of interest in the top-view image (Figure 7 ( the shaded area of a).

As shown in Figure 6, it is the circumferential image of the fruit 7 with cross laser markings (shown in Figure 6 (a)) and the fruit without cross laser markings collected by one of the circumferential cameras 2 Circumferential image (shown in panel (b) in Figure 6). The dashed line in (b) in Figure 6 is the line formed by the laser. Subtract the circumferential image of the fruit with cross laser markings (shown in Figure 6 (a)) and the fruit circumferential image without cross laser markings (shown in Figure 6 (b)) to get the circumferential to the laser line image. Multiply the lower area of the peripheral laser line image by the peripheral image of the fruit without the cross laser marking (shown in (b) in Figure 6) to obtain the region of interest in the peripheral image (Figure 7 (b) shaded area).

As shown in Figure 8, the ROI of the top-view image and the 4 ROIs of the circumferential image are stitched together to obtain an image without redundant information on the surface of the spherical fruit.

The groove photoelectric sensor 8 that the present invention adopts is FGL 220-R-PSM3; Microprocessor U1 is AT89C2051; Optocoupler U2 and optocoupler U5 are 4N25; Optocoupler U3 is AIA UDK-0 5Vdc; Motor driver U4 is Fuji RYHD751 - FS-VV2; relay U6 is SRD-5VDC-SL-C; transistor Q1 is 8050; diode D1 is 1N4148.

Claims (3)

1. for a method for the irredundant image information acquisition of spherical fruit, it is characterized in that: with 1 top view camera and 4 circumferential cameras, fruit top view picture not with cross laser graticule of the fruit circumference image of the fruit top view picture of 1 web cross laser graticule and 4 web cross laser graticules, 1 width and the fruit circumference image of 4 width not with cross laser graticule are gathered respectively to the fruit in visual field; By fruit top view picture not with cross laser graticule of the fruit of the fruit top view picture of 1 web cross laser graticule and 4 web cross laser graticules circumference image and 1 width and the fruit circumference of 4 width not with cross laser graticule, image is corresponding respectively subtracts each other, and obtains 1 width top view laser line image and the circumferential laser line image of 4 width; The border circular areas of top view laser line image is multiplied with the fruit top view picture not with cross laser graticule, obtains top view interesting image regions; Respectively 4 width circumference laser line image are multiplied with the corresponding fruit not with cross laser graticule circumference image, obtain 4 width circumference interesting image regions; By top view interesting image regions and circumferential interesting image regions split, obtain the irredundant frame of fruit surface.

2. a kind of device for the irredundant image information acquisition of spherical fruit of method as claimed in claim 1, comprise 4 circumferential cameras (2), top view camera (3), trigger controller (4), computing machine (5), flute profile photoelectric sensor (8) and travelling belt (10), is characterized in that: also comprise servomotor (1), 4 cross laser line marking devices (6) and pallet (9);

4 circumferential cameras (2) are evenly arranged in same plane ringwise, 4 cross laser line marking devices (6) are evenly arranged between 4 circumferential cameras (2), the driving-belt (10) driven by servomotor (1) is arranged in circumferential camera (2) and cross laser line marking device (6) below, the space angle 22.5 degree of driving-belt (10) and a circumferential camera (2); Servomotor (1) drives driving-belt (10) to rotate, fruit (7) is loaded on pallet (9), pallet (9) is arranged on driving-belt (10), installs top view camera (3) in fruit (7) top; Flute profile photoelectric sensor (8) across driving-belt (10) both sides, and is positioned at pallet (9) working direction front;

Trigger controller (4) respectively with servomotor (1), flute profile photoelectric sensor (8), 4 circumferential cameras (2) are connected with top view camera (3), and computing machine (5) is connected with top view camera (3) with 4 circumferential cameras (2) respectively.

3. a kind of device for the irredundant image information acquisition of spherical fruit according to claim 2, is characterized in that, described trigger controller (4): comprise microprocessor U1, optocoupler U2, optocoupler U3, motor driver U4, optocoupler U5, relay U6, triode Q1, diode D1, pot Rp, resistance R1, resistance R2, resistance R3, resistance R4, resistance R5 and resistance R6;

Brown lead-in wire and the blue lead-in wire of flute profile photoelectric sensor (8) meet+24V power supply and GND1 respectively, flute profile photoelectric sensor (8) black lead connects optocoupler U2 the 1st pin by resistance R2, optocoupler U2 the 2nd pin meets GND1, resistance R1 is connected on 6 pins of+5V power supply and microprocessor U1, optocoupler U2 the 5th pin connects microprocessor U1 the 6th pin, optocoupler U2 the 4th pin meets GND0, microprocessor U1 the 19th pin connects camera trigger pip input end, resistance R4 connects+5V power supply and optocoupler U5 the 1st pin respectively, optocoupler U5 the 2nd pin connects microprocessor U1 the 17th pin, optocoupler U5 the 5th pin connects+24V power supply, resistance R5 is connected between optocoupler U5 the 4th pin and triode Q1 base stage, relay U6 coil is connected between+24V power supply and triode Q1 collector, transistor emitter meets GND1, diode D1 reversal connection is at the coil two ends of relay U6, the normally opened contact of relay U6 is connected between+3V power supply and 4 cross laser line marking devices (6), the power cathode of 4 cross laser line marking devices (6) meets GND1, optocoupler U3 the 1st pin connects microprocessor U1 the 18th pin, optocoupler U3 the 2nd pin meets GND0 by resistance R7, optocoupler U3 the 3rd pin connects+24V power supply, optocoupler U3 the 4th pin meets the 2nd pin and the resistance R6 of the CN1 interface of motor driver U4 respectively, 13rd pin of the CN1 interface of motor driver U4 meets GND1 and resistance R6, resistance R3 is connected between the 22nd pin of the CN1 interface of+24V power supply and motor driver U4, pot Rp meets the 22nd pin and GND1 of the CN1 interface of motor driver U4, the TB3 interface of motor driver U4 connects servomotor (1).