Transactions of the Chinese Society of Agricultural Engineering, the 1st in Agricultural Engineering, is supervised by China Association for Science and Technology, and sponsored by Chinese Society of Agricultural Engineering. It aims to introduce the latest scientific achievements and developing trends of Agricultural Engineering and provides the academic developments abroad and domestic of the discipline. The scope covers agricultural water-soil engineering, agricultural information and electrical technology, agricultural products processing engineering.
The journal is included in EI, JST, Pж(AJ), CA and CSCD.
Editor-in-Chief Zhu Ming
Deputy Editor-in-Chief Wei Xiuju Zhang Ruihong Xi Weimin Wang Liu Wang Yingkuan Li Pingping Ying Yibin Tong Jin Yun Wenju Zhao Chunjiang Kang Shaozhong
The timely and accurate crop distribution maps derived from satellite observations could assist in crop growth monitoring. Although crop mapping methodologies have been widely studied, there are still some drawbacks, such as the limitation of ground reference data and low efficiency of crop type mapping caused by using time series data of the entire year. The objectives of this study were as follows: (1) to develop a new method, which could identify crop types using the crop records of the previous years; (2) to evaluate the performance of the method with different time series length, and try to acquire the crop type map at 30 m spatial resolution. The study area of this paper was the ASD30 of Kansas State, US. We firstly used the cropland data layer (CDL) data and MODIS EVI (enhanced vegetation index) time series between 2006 and 2013 to generate reference EVI time series with the ABNet algorithm for the major crops in the study area,
i.e., alfalfa, corn, sorghum, and winter wheat. Then, we acquired the “possible” training samples in 2014 using the CDL records between 2006 and 2013. If a pixel was labeled as “Crop A” more than 4 times among the 8-year CDL records, the pixel was labeled as “possible Crop A” in 2014. Next, we compared the MODIS EVI of the “possible crop A” pixels and the reference EVI time series of Crop A. If the two profiles were matched, the “possible Crop A” was confirmed as a training sample of “Crop A”. Finally, we used these training samples and monthly composited Landsat NDVI (normalized differential vegetation index) to identify crop types at 30 m resolution. To analyze the effect of time series length on crop type recognition performance, we tried seven time series lengths (April, April–May, April–June, April–July, April–August, April–September, and April–October), used MODIS EVI time series to acquire training samples for each time series length, and then identified crop types using the corresponding training samples and Landsat NDVI time series. Several metrics derived from the confusion matrix, such as overall accuracy, Kappa coefficient, were used to evaluate the classification performance. The results showed that when only the time series data in April were used, we acquired 5 088 samples, and 91.86% among these samples had the same crop label with the CDL data. When longer time series data were used, more training samples in 2014 were acquired with high accuracy. When the entire EVI time series data were applied, 10 803 samples were acquired and 10 317 samples had same crop label with CDL data. When these training samples and monthly composted Landsat NDVI were used to identify crop types at 30 m resolution, the classification accuracy was low if the April or April–May time series data were used, and the overall accuracy were 66.12% and 52.51%, respectively. When the time series length was April–October, the overall classification was 94.89%. The April–August time series achieved good classification performance, as 10 183 training samples were acquired, 96.32% samples had same label to CDL data, the overall classification accuracy was 94.02%, and the acreage of major crops was similar to CDL data. Finally, we can get the following conclusions: (1) The method proposed in this study can acquire train samples in the classification year when the ground reference data are absent. Using these training samples, we can obtain crop type distribution maps with high accuracy (better than 90%). (2) We can acquire the crop type map of the study area in August with the high classification accuracy which is similar to the result derived from the entire EVI time series, and has the similar crop acreage with the CDL data for each crop. In the future, we can enhance this method by improving the previous-year training samples with the CDL crop confidence layer.
The phenotypic traits of corn earS are important quantitative data in maize breeding and variety identification. In tradition, breeders are employed to deal with lots of corn ears by means of manual measurement and visual count. However, this process requires time and effort, and the measured traits are prone to be subjective and incomplete. In recent years, based on machine vision and image analysis, some semi-automatic systems have been developed and applied to the maize variety test. However, ful automation is still a challenge owing to the strict high-throughput and high-precision requirements in large-scale maize breeding. To balance efficiency and accuracy of variety test for corn ears, a high-throughput phenotypic measurement method and system based on panoramic surface image was proposed. Firstly, a novel mechanic system was proposed, which automatically conveyed the corn ears above a chain-roller structure, while the rolling corn ears were continuously imaged by a fixed industrial camera that was perpendicular to the moving plane of corn ears. In only several seconds, dozens of side images in which corn ears were in different positions could be collected to generate the image dataset of single corn ears. By analyzing the movement state of corn ears, a transformation model which describes the relationship among ear roll, camera imaging and surface position was then built to bridge the image sequence and the panoramic surface image of corn ears. The corn ears in the image sequence were respectively segmented and the center axes were dynamically determined by figuring out the shape and bounding box. This model always extracted the most appropriate subregions of corn ear from image sequence, and then stitched them to the calculated positions on the panoramic surface image. As a result, the panoramic image of corn ear demonstrated the three-dimensional surface information in a two-dimensional image, and thus provided a more intuitive and complete way for phenotyping calculation of corn ears. The valid surface region of corn ears in the panoramic image was further determined by the boundary detection technique that was performed by evaluating the perimeters of corn ears in the image sequence. Robust kernel segmentation based on hierarchical threshold method was also utilized to extract all candidate kernels which satisfied the area and shape constraints, and some more restrictive filters based on machine learning methods, such as SVM (support vector machine), could also be taken to evaluate the validation of kernels. The segmented kernels in the panoramic image were used to calculate the total kernels, rows per ear and kernels per row. The experimental results showed that the proposed method and system could achieve optimized efficiency and accuracy balance. High-throughput convey mechanism improved the efficiency of image acquisition to 15 ears per minute. Compared with the methods based on single and multiple images, the variety test method based on panoramic surface image scan make full use of the entire surface information of corn ears and reveal its individual phenotypic traits. The computation accuracy of ear length, ear diameter, rows per ear, kernels per row and total kernels was up to 99%, 91.84%, 97.15%, 98.89%, and 95.37%, respectively.
In order to solve the problem that the seedlings cultivated with the pie-shaped compressed substrate could not be planted with the existing transplanter, a semi-automatic transplanter was designed for the seedlings cultivated on compressed substrate in this paper by mimicking the artificially transplanting method of putting seedlings after punching a hole. The transplanter mainly consisted of a ground wheel, a swing mechanism, a ratchet wheel, a hole puncher, a shifting mechanism, a seedling clamp mechanism, a conveying device, a driving system, and a rack. The physical dimensions and mechanical properties of the seedlings cultured on compressed substrate were the key basis for the design of the seedling planting schemes and structures. Taking watermelon seedlings as the research object, the dimensions of the seedlings were measured, and the friction coefficients of the compressed substrates with different water contents and the compressive strength of the compressed substrate was determined. The coefficients of friction between the slides and the sides of watermelon seedlings with two groups of different water content were determined to be 0.755 and 0.634, respectively, by single-factor tests. The relationship between the compression load and the compression amount of the two groups was also analyzed. When the load was 0–5 N, as the surface of the seedling was in point contact with the semi-circular thin metal at the initial stage of compression, the compression load increased evenly and the amount increased rapidly. When the load was between 5 and 20 N, as the surface of the seedling was in surface contact with the semi-circular thin metal, the compression load increased evenly with less impact on compression. When the load was greater than 25 N, some external cracks were observed on the surface of the seedlings during the test. The comparison of the two tests showed that the seedling with high water content was not easily destroyed. The swing mechanism was optimized according to the known movement law of the initial angle of the crank and the output angle of the driven rod. The dimensions of optimized parts were 57, 161, 79, and 170 mm, respectively. When the crank rotated one revolution of 360°, the reciprocating swing angle of the driven rod was 92.3°, which satisfied the working requirements of the four-equal-part ratchet wheel mechanism. The structure and specific size parameters of the hole puncher were determined according to the measurement size and planting depth requirements of the watermelon seedling. The width of the edge surface was 1.5 mm. The inner diameter of the small end was 64 mm, and the height of the tapered part of the hole puncher was 65 mm; when the incision angle was 21°, the inner diameter of the big end was about 114 mm. The structure parameters of the seedling clamp mechanism were determined according to the mechanical properties and the dimension of watermelon seedlings. It was also concluded that the seedling’s matrix could overcome the self-gravity of the seedling, which ensured that the seedling clamp mechanism could securely grip the seedlings when transplanting with a clamping force of 26 N. The seedling conveying device and the planting holding device were driven by the same power source, ensuring that the feeding speed of the seedlings was synchronized with taking the seedling action of the planting holding device. The conveyor belt was used to transport the seedlings, and the conveyor belt was designed to send seedlings at a speed of 40 plants per minute. Using a compressed substrate for field planting functional verification tests, the average plant spacing was 98.6 cm when the transplanter moved at a stable speed of 2.1–2.6 km/h. The pass rate of the plant spacing was 90.62% and the lodging rate was 21.9% which was slightly higher. In the follow-up study, dual-ground-wheel driving would be used to improve the reliability of the transmission and obtain uniform spacing; a hole shape with the same level and the same depth should be acquired, and the lodging rate after the landing of pie-shaped matrix would be reduced by increasing the copying mechanism and optimizing the shape and structural parameters of the hole puncher.
