Yi Xun

Multi-template matching algorithm for cucumber recognition in natural environment

A multi-template matching library was established by proportional scaling and rotating the standard cucumber image.Multi-template matching algorithm for cucumber recognition for harvesting robot was developed.High recognition accuracy of 98% was obtained from the verification test. The automatic recognition of cucumber target within its cultivating environment is one of the key techniques for the cucumber harvesting robot. Since cucumber grows in the complex environment and its color is similar to that of branches and leaves, it is quite challenging to achieve high identification accuracy when employing algorithms based on color features, image segmentation and shape features. Adopting spectroscopy can simplify the algorithm. However it increases the complexity and cost of the robot system. The multi-template matching method was proposed to solve this problem in this paper. A multi-template matching library, which contained 65 cucumber images, was established based on the statistical parameters of the matured Radit cucumber, by proportional scaling the standard cucumber image with step of 0.1 in the range of 0.8, 1.2 and rotating with step of pi/36 in the range of -pi/6, pi/6. To identify the cucumber in the visual field of the robot, cucumber templates in the library are used to calculate the matrix of normalized correlation coefficients (NCC) with the target image, one after another. If the maximum NCC is above the threshold, there is the target cucumber in the image frame. Otherwise, there is no target in the visual field. To verify the algorithm, 100 photos of the Radit cucumber with different size and angle were sampled in the test. The results indicated that cucumbers were correctly recognized and positioned in 87 images. Cucumbers which were correctly recognized but with picking position deviation appeared in 11 images. Cucumbers were not found in two images. In general, the correct recognition accuracy is 98%, with 11% fault position.

Navigation line detection based on robotic vision in natural vegetation-embraced environment

In order to detect navigation line for agricultural vehicles in natural vegetation-embraced environment, a method was presented based on robotic vision. Firstly, an improved region growing algorithm was introduced to segment path. Then the two edges of path were extracted and an array of points through the center was computed. Lastly, the guidance line for vehicles to follow was obtained by Hough transform. The method was also extended to forestry environments and tall-crop fields where planting is discretely spaced. Difference was that the two edges of a path were square-fitted to offset edge imperfections. Batch processing of images under Matlab shows logicalness and steadiness of this method and its extensive appliance in various scenarios.

A Dynamic Threshold Segmentation Algorithm for Cucumber Identification in Greenhouse

Aiming at realizing cucumber identification and location for cucumber harvesting robot in greenhouse, a segmentation algorithm for cucumber image was presented. Using M (M=2G) component as the threshold segmentation channel, choosing an initial threshold to segment the images. The threshold was dynamic revised based on shape characteristics of cucumber fruits after the initial segmentation threshold was judged. The pixel region which did not belong to fruit was removed by erosion and dilation using structure elements. Final segmentation images were obtained after region marking. Thirty images were tested to verify the correctness of the algorithm, and twenty-three images were segmented correctly with satisfactory results. Experimental results show that the segmentation success rate is 76.7%, and the segmentation effect is acceptable which can be used in greenhouse cucumber identification.

Sliding-mode position control of robot joint based on self-adaptive parameters adjusting

The single robot joint, which contains a DC torque motor and a harmonic reducer, was designed. And sliding mode control strategy by the self-adaptive adjustment of the sliding parameter ε was proposed in this paper. This control strategy can adaptively adjust ε by switching variables and their derivatives according to the system. It is applied in the control of robot joint to solve the contradiction between the exponential reaching law response speed and chattering. Simulation results show that the open-loop control response time of the parameter adaptive adjustment sliding mode control is 0.3s. The experiment results show that under the control of adaptive sliding mode with joint angle 45 and rise-time 0.23s, the maximum overshoot is 0.4%, with no chattering and a good dynamic quality.

Recognition of Broken Corn Seeds Based on Contour Curvature

Aiming at the problem on determining breakage rate of corn seeds in real time, a method for identifying broken corn seeds was presented based on contour curvature. The broken corn seeds were classified into three classes according to damage position and level. Firstly, the seed contour was rebuilt with Fourier descriptors to achieve smooth contour. Then, contour curvature was calculated and the point with maximum curvature value on the contour was taken as an endpoint of the seed major axis, so the position of major axis was determined. Two kinds of broken seeds were identified by judging their shape symmetry. Lastly, other broken seeds were identified using two parameters. One parameter was the difference between maximum and sub- maximum peaks of curvature. The other was the frequency of the curvature line crossing the mean line. The experimental results on three corn varieties show that the average recognition rate of broken corn seeds is 92.4% and the method is valuable for engineering application.