Search

Copyright of Transactions of the Chinese Society of Agricultural Engineering is the property of Chinese Society of Agricultural Engineering and its content may not be copied or emailed to multiple sites or posted to a listserv without the copyright holder's express written permission. However, users may print, download, or email articles for individual use. This abstract may be abridged. No warranty is given about the accuracy of the copy. Users should refer to the original published version of the material for the full abstract. (Copyright applies to all Abstracts.)

Copyright of Transactions of the Chinese Society of Agricultural Engineering is the property of Chinese Society of Agricultural Engineering and its content may not be copied or emailed to multiple sites or posted to a listserv without the copyright holder's express written permission. However, users may print, download, or email articles for individual use. This abstract may be abridged. No warranty is given about the accuracy of the copy. Users should refer to the original published version of the material for the full abstract. (Copyright applies to all Abstracts.)

The navel orange planting area in Gannan region of China has reached the first in the world, with an annual output of one million tons. However, most navel oranges were picked manually with high labor intensity and low efficiency. It was very urgent to introduce intelligent picking robots. In the process of navel orange picking, the picking robot needed to obtain the spatial position and fruit size information of navel orange fruit in real time to achieve efficient and rapid intelligent grading picking. This study proposed an algorithm framework--OrangePointSge based on the improved YOLACT model and the least-square sphere fitting method. The real-time instance segmentation algorithm YOLACT generated an instance mask and cropped the registered depth image to obtain the fruit depth point cloud. Then, the least-square method was used to fit the navel orange fruit shape, and the centroid coordinates and radius under the camera coordinate system were obtained. Firstly, the RGB-D data of navel orange fruit was collected by Microsoft's latest consumer-level depth camera (Azure Kinect DK), and the navel orange fruit instance segmentation dataset and enhancement dataset were established and free to open source. The dataset contained 2 178 images, including 8 712 samples. Among those samples, the number of non-occluded fruit was 4 682, and the number of slightly occluded fruit was 4 030. Then the YOLACT algorithm was improved to adapt to the detection of navel orange fruit, and the instance mask of the fruit was output. We modified the original ResNet (deep residual network) backbone network to HRNet (High-Resolution Net) to simplify the model, and used the hrnet_w48 structure for feature extraction to improve the model detection accuracy. We also optimized the non-maximum suppression process to improve the detection speed. At the same time, several groups of comparative tests were also set up, including the performance of YOLACT with different backbone networks and the performance of different algorithms in different scenarios. To improve the generalization ability of the training model, we carried out a series of enhancement algorithms for color images, mainly including scale change, color change, brightness change, and adding noise. For long-range small targets, we used oversampling method for enhancement training to improve their detection accuracy. Therefore, the recognition speed and mask accuracy of navel orange were higher than other algorithms. The detection speed was 44.63 frames/s and the average accuracy was 31.15%. The mask generated by improved YOLACT will be cropped with the depth image registered with the color image to obtain the depth information of each fruit to generate the fruit depth point cloud from the depth image. The least-square method was used to fit the point cloud to obtain the centroid coordinates and radius of the fruit, which were used to guide the robot to carry out grading picking. In addition, we added the comparison with RANSAC (Random Sample Consensus) algorithm. The results show that the least-square method is faster than RANSAC while ensuring the appropriate accuracy. When the number of point clouds was 1 400-2 000, the fitting time was 1.99 ms, the positioning error was 0.49 cm, the fitting radius root mean square error was 0.43 cm, and the volume root mean square error was 52.6 mL. Through the experiment of setting different distances on the fitting accuracy, it was found that when the number of point clouds was more than 800 and the distance was less than 1.0 m, the positioning error was controlled within 0.46 cm. Finally, by introducing parallel computing and combining the above two processing flows, the overall processing speed of OrangePointSge was 29.4 frames/s, which can better balance the accuracy and speed. Thus, the proposed algorithm is conducive to practical application and engineering deployment. [ABSTRACT FROM AUTHOR]

Copyright of Transactions of the Chinese Society of Agricultural Engineering is the property of Chinese Society of Agricultural Engineering and its content may not be copied or emailed to multiple sites or posted to a listserv without the copyright holder's express written permission. However, users may print, download, or email articles for individual use. This abstract may be abridged. No warranty is given about the accuracy of the copy. Users should refer to the original published version of the material for the full abstract. (Copyright applies to all Abstracts.)

Cracks are common in nature and affect a number of soil properties. They increase the permeability, reduce the soil strength and stability, and thus affect the growth of plant roots. This paper presented a model for crack formation which was introducing the correlation length to constructing the soil critical strain random field to optimize the cracking model. The model was based on a lattice of Hookean springs with finite strength and the critical strain. The concept of correlation length was introduced to constructs the critical strain random field, the correlation length was used for reflects the spatial continuity of the soil. The critical strain random fields were generated using program written by Fortran language. The model reproduced the dynamics characteristic of the development of crack network in nature, and some of the parameters of the model represented the physical properties of the soil. We processed the digital images to quantitatively analyze the geometric characteristics of cracks. The typical procedure of digital image processing included: the color image of the crack patterns was converted into a gray one, the gray image was further changed into a binary one with the threshold division method. After the process, the cracks and aggregates were simply distinguished in a black-white image, in which the black lines and the white areas represent the crack networks and the aggregates, respectively. We optimized the threshold to make the binary images as clear as possible, and processed the images through binarization, noise removal and edge extraction of crack for extraction of morphological properties. Finally, the influence on model simulation cracking of spatial correlation length, heterogeneity in random field was analyzed. Comparison of crack area, crack length, Euler number density by calculating experiment and simulation images, it could be drawn that crack morphology could be described by Minkowski area, length, Euler number density effectively. By calculating the average μ, standard deviation δ of cracks area, length, Euler number density, the images of average density value μ and μ±δ are basically the same in the change trend between experimental and simulated process. After calculation, the coefficients of determination R2 for the area, length and Euler number density are all larger than 0.5, and the root mean square error and deviation are small enough. The consistency index IA quite approach to 1, which indicating that the calibration results are satisfactory. The two models are compared, among them the determination coefficient R2 is between 0.889-0.968, the consistency index IA is between 0.964-0.993, which reflecting the dynamic cracking process of cracks well. The greater the correlation length and the smaller the variance, the lower the discretization degree of the random field, that the more uniform the soil cracking. On the contrary, there are mainly isolated cracks. And the mean value is too small, the whole strain of the random field is low, the simulated crack does not consistent with the development of the natural crack. [ABSTRACT FROM AUTHOR]

Preferential flow refers to such a flow transport pattern that water and solute flow round soil matrix, which accelerates soil water movement. Soil cracks are a part of reasons that lead to preferential flow. This study aimed to analyze soil preferential flow pattern in order to establish a model of preferential model based on soil fracture network. A field color tracer experiment was carried out in Ningxia University from July to September in 2015 under precipitation intensity of 20 and 50 mm/h, and rainfall duration of 20-60 min. The results showed that when the rainfall intensity was 20 mm/h, the water infiltration depth increased with the increase of rainfall duration and the preferential flow began to develop, but it was still overall-matrix flow and the preferential flow was only locally developed. When precipitation intensity was 50mm/h and the duration of rainfall was 20min, water flows bypassed the soil matrix and preferential flow phenomena was observed obviously. The infiltration depth exceeded the fracture depth after duration of rainfall increased to 50min, soil fracture had not play a major role in the soil water movement, and the flow patterns were overall-matrix flow. The results showed that the flow patterns were mainly matrix flow under lower rainfall or irrigation intensity, and preferential flow under higher rainfall or irrigation intensity. A coupling model of soil matrix and fracture network for simulating preferential flow was established (CP model). In this model, the soil medium was divided into matrix and fracture networks formed by flow channels, each having its own saturated hydraulic conductivity and water movement equation. These equations were solved by using the control volume finite element method, and the nonlinear equations were linearized by Newton-Raphson methods to improve the convergence, and shorten the operation time. Equation of water movement in matrix and fracture network flow equation were coupled using global approach which solved the problems of the flow mutual influence caused by the cross-fracture network, and improved the computational efficiency. The validity of the proposed model was tested by the analysis of the soil moisture content between the simulated and the observed values. The results showed that the root mean squared error (RMSE) of soil moisture was 0.016-0.025, R2 of soil moisture was above 0.78, bias was 0.020-0.026, index of agreement (IA) was 0.918-0.938, indicating good simulation by the CP model. In the end, different crack structure but statistically equivalent networks was generated to validate the model simulation and produced well simulation results. These numerous fractures had stable effect on the variances of soil moisture. The results of this paper can provide valuable information for analyzing preferential flow and the coupling mechanism of the water content and solute transport between soil dry shrinkage crack and soil matrix. [ABSTRACT FROM AUTHOR]

Soil cracking is a common phenomenon in nature, which has great influence on soil properties, plant growth and soil moisture transfer. This study aimed to reveal the development law and formation mechanism of dry shrinkage cracks on the soil surface. Soil crack evolution test was carried out in 1-m2 soil in Yellow River irrigation area (105°40'E, 37°28'N) of Ningxia Zhongning County. Soil was irrigated to saturate condition, and dried under sunlight to form cracks. A Canon digital camera was used to take crack pictures every 2 h for 3 days. The picture was processed through binarization, noise removal and edge extraction of fracture for extraction of morphological properties. Minkowski area, length and Euler number were to quantify crackers. The cracker evolution was simulated through morphological dilation and erosion operation. The model assumed that the soil critical strain field followed normal distribution. Friction was taken as soil static adhesion force, and grid structure contraction frequency represented soil moisture evaporation intensity in hexagonal network structure composed of nodes. In the simulation, vertical interaction between layers of the soil was ignored, and the clay was considered as elastic material. The distance between each node in hexagonal structural network was reduced gradually on behalf of the soil evaporation. The reducing of distance between each node could cause the changing of force between network nodes, which resulted in soil cracking. The model was solved using program written by C++ language under the CodeLite platform and generated Tiff image as output. By comparison of crack area, crack length, euler number density that calculated by analyzing experiment pictures and images the program produced, it could be drawn that crack morphology could be described by Minkowski area, length, euler number density effectively; each parameter had its own variation as morphological analysis results did, among which area density function was an increasing function between 0 and 1, the length density function increased and decreased, and the euler number density function increased and then decreased, and finally increased again. Furthermore, a two-dimensional soil cracking model was established based on Hooke's law. To analyze the calculation results of different soil cracks, statistically equivalent soil cracks had been investigated, 100 seeds were used to generate 100 groups of random structured soil cracks. The average level and standard deviation of each parameter of 100 groups of random structured soil cracks were also calculated and the average density value in this experiment was basically distributed between the mean value and standard deviation of the model. Coefficient of determination of area, length and euler number density was 0.893-0.928. Root mean square error was 0.002-0.039. Bias was 0.064- 0.144. Index of agreement was greater than 0.888. It suggested that the model established had high precision. Therefore, the model based on Hooke's law can effectively simulate the cracking process of soil. Simulation of surface soil cracking with numerical method is helpful to study the formation of farmland soil cracks as well as its dynamic evolution mechanism. [ABSTRACT FROM AUTHOR]