| Removal of aillary bud is the key of increase yield of cherry tomato. Because the color feature of aillary bud is similar to the surrounding branches and leaves. In order to successfully remove axillary buds of Tomato, study the image acquisition removal of point positioning, lateral horizontal projection angle measurement and some other problems. Research contents and results:1. Study on the agronomic technology, using the method of semi cut with cherry tomato bud. Measure the geometric characteristics of Cherry tomato, and mechanical properties of the axillary bud and lateral branch. Get the data of the plant height, stem diameter, pitch, lateral orientation and angle. Implement standardized cultivation in green house. In accordance with the row spacing, plant spacing, spacing requirements for fixed beds.2. In this paper, an image processing method based on blue light staining is proposed. Use monocular camera which captures images and ultrasonic displacement sensor to get the 3D coordinate point of axillary bud growth point. It is difficult to segment image, because the color of the background is same to the axillary buds, branches and stem of Cherry Tomato. Using the blue LED light source to irradiate the part of axillary buds can dye the buds blue. The background is the other tomato plants which the color is still green, so it’s easy to extract the object from image. When the distance between the LED light source and the plant is 13cm, the collected image is the most complete. Take the B component images in RGB spatial domain.3. B component image is gray image and its histogram is bimodal. Select gray value in the trough as a threshold, and then segment image, the outline of the object can be got clearly. But there are burrs on the edge of the outline, so the gray image should be translate into frequency-domain diagram by fast Fourier transform (FFT). Use the low pass filter to filter out the burrs at high frequency, and retain the outline at low frequency. Pass rate is set to 2.8% of the maximum frequency of the image. In order to highlight the characteristics of the key feature points, use morphological dilation to process the image by 7×7 cross structure element. Then find out all the corners on the image by the Shi-Tomasi corner detection algorithm. Analysis the growth characteristics of cherry tomato axillary buds, and then set a discriminant condition. Iterate over all the corners, if there are two corners in accordance with the discriminant condition, then the two points are the key feature points. The mid-point of the two points is the axillary bud growth point. Identify 90 images with axillary bud growth. Identify 42 images without axillary bud growth, correct recognition rate is 100%.82 images can be detect the axillary bud successfully, correct recognition rate is 94.4%. After the removal of axillary buds, the removal success rate is88.9%.4. When using automatic robot remove the cherry tomato axillary buds, need to get the horizontal projection direction angle of collateral for adjusting the position and angle of the robot end effector. So the camera can capture the projection plane image of stems, axillary buds and branches, and perform the following steps of identifying axillary buds by machine vision. In this paper, design a photoelectric automatic rotating direction finding mechanism. Through the telescopic rack control the whole direction-finding mechanism extending and retracting. Close the two semi-rings of floating ring by mechanical claw, make the two semi-rings surround the main stem, and then make them move upward by manipulator control. The inner diameter of the floating ring is 60mm, the floating ring rotated 90 degrees on the track groove recurrently. The direction-finding mechanism detect lateral branch by 8 photoelectric sensors which have been fixed on floating ring, and get the horizontal projection angle of tomato lateral branches by position angle of sensor on floating ring and the rotation angle of the floating ring. The experiment shows that:when the proportion coefficient is 1.5, the angular velocity is 1.875π/s, line speed is 2.5cm/s, detection success rate is 97%, the standard angle difference is ±8.6°. Altitude compensation is set to 20mm, the success rate is 91%. Jump height is set to 10mm, the success rate is 89.7%. The success ratio of the test to 42 cherry tomato plants was 88.2% in the 1.8m altitude.5. The industrial camera is calibrated by the black and white chess board, and the result of the calibration error is within 2mm. Analysis the shortage of monocular camera ranging, positioning system using ultrasonic ranging sensors and cameras, and calibration of the ultrasonic ranging sensor, according to the relative camera coordinate system of beam vector.6. Complete the design and assembly of axillary bud removal of the robotic end effector. The mechanical navigation method based on the guide rail was determined, and the position of the cherry tomato plants was accurately located by using the two positioning method. The forward and inverse kinematics of the manipulator is analyzed. The total successful rate was 83%. |
PDF Full Text Request