Crop rows detection in maize fields remains a challenging problem due to variation in illumination and weeds interference under field conditions. This study proposed an algorithm for detecting crop rows based on adaptive multi-region of interest (multi-ROI). First, the image was segmented into crop and soil and divided into several horizontally labeled strips. Feature points were located in the first image strip and initial ROI was determined. Then, the ROI window was shifted upward. For the next image strip, the operations for the previous strip were repeated until multiple ROIs were obtained. Finally, the least square method was carried out to extract navigation lines and detection lines in multi-ROI. The detection accuracy of the method was 95.3%. The average computation time was 240.8 ms. The results suggest that the proposed method has generally favorable performance and can meet the real-time and accuracy requirements for field navigation. Keywords: machine vision, crop rows detection, navigation, multi-ROI DOI: 10.25165/j.ijabe.20211404.6315 Citation: Zhou Y, Yang Y, Zhang B L, Wen X, Yue X, Chen L Q. Autonomous detection of crop rows based on adaptive multi-ROI in maize fields. Int J Agric & Biol Eng, 2021; 14(4): 217–225.

As the basic link of autonomous navigation in agriculture, crop row detection is vital to achieve accurate detection of crop rows for autonomous navigation. Machine vision algorithms are easily affected by factors such as changes in field lighting and weather conditions, and the majority of machine vision algorithms detect early periods of crops, but it is challenging to detect crop rows under high sheltering pressure in the middle and late periods. In this paper, a crop row detection algorithm based on LiDAR is proposed that is aimed at the middle and late crop periods, which has a good effect compared with the conventional machine vision algorithm. The algorithm proposed the following three steps: point cloud preprocessing, feature point extraction, and crop row centerline detection. Firstly, dividing the horizontal strips equally, the improved K-means algorithm and the prior information of the previous horizontal strip are utilized to obtain the candidate points of the current horizontal strip, then the candidate points information is used to filter and extract the feature points in accordance with the corresponding threshold, and finally, the least squares method is used to fit the crop row centerlines. The experimental results show that the algorithm can detect the centerlines of crop rows in the middle and late periods of maize under the high sheltering environment. In the middle period, the average correct extraction rate of maize row centerlines was 95.1%, and the average processing time was 0.181 s; in the late period, the average correct extraction rate of maize row centerlines was 87.3%, and the average processing time was 0.195 s. At the same time, it also demonstrates accuracy and superiority of the algorithm over the machine vision algorithm, which can provide a solid foundation for autonomous navigation in agriculture.

Most of the current crop row detection algorithms focus on extracting crop canopy rows as location information. However, for some high-pole crops, due to the transverse deviation of the position of the canopy and roots, the agricultural machinery can easily cause the wheel to crush the crop when it is automatically driven. In fact, it is more accurate to use the crop root row as the feature for its location calibration, so a method of crop root row detection is proposed in this paper. Firstly, the ROI (region of interest) of the crop canopy is extracted by a semantic segmentation algorithm, then crop canopy row detection lines are extracted by the horizontal strip division and the midpoint clustering method within the ROI. Next, the Crop Root Representation Learning Model learns the Representation of the crop canopy row and crop root row to obtain the Alignment Equation. Finally, the crop canopy row detection lines are modified according to the Alignment Equation parameters to obtain crop root row detection lines. The average processing time of a single frame image (960 × 540 pix) is 30.49 ms, and the accuracy is 97.1%. The research has important guiding significance for the intelligent navigation, tilling, and fertilization operation of agricultural machinery.

Machine vision-based navigation in the maize field is significant for intelligent agriculture. Therefore, precision detection of the tasseled crop rows for navigation of agricultural machinery with an accurate and fast method remains an open question. In this article, we propose a new crop rows detection method at the tasseling stage of maize fields for agrarian machinery navigation. The whole work is achieved mainly through image augment and feature point extraction by micro-region of interest (micro-ROI). In the proposed method, we first augment the distinction between the tassels and background by the logarithmic transformation in RGB color space, and then the image is transformed to hue-saturation-value (HSV) space to extract the tassels. Second, the ROI is approximately selected and updated using the bounding box until the multiple-region of interest (multi-ROI) is determined. We further propose a feature points extraction method based on micro-ROI and the feature points are used to calculate the crop rows detection lines. Finally, the bisector of the acute angle formed by the two detection lines is used as the field navigation line. The experimental results show that the algorithm proposed has good robustness and can accurately detect crop rows. Compared with other existing methods, our method's accuracy and real-time performance have improved by about 5 and 62.3%, respectively, which can meet the accuracy and real-time requirements of agricultural vehicles' navigation in maize fields.

Real-time crop harvest data acquisition from harvesters during harvesting operations is an important way to understand the distribution of crop harvest in the field. Most real-time monitoring systems for grain yield using sensors are vulnerable to factors such as low accuracy and low real-time performance. To address this phenomenon, a real-time grain yield monitoring system was designed in this study. The real-time monitoring of yield was accomplished by adding three pairs of photoelectric sensors to the elevator of the corn kernel harvester. The system mainly consists of a signal acquisition and processing module, a positioning module and a visualization terminal; the signal acquisition frequency was set to 1 kHz and the response time was 2 ms. When the system operated, the signal acquisition and processing module detected the sensor signal duration of grain blocking the scrapers of the grain elevator in real-time and used the low-potential signal-based corn grain yield calculation model constructed in this study to complete the real-time yield measurement. The results of the bench tests, conducted under several different operating conditions with the simulated elevator test bench built, showed that the error of the system measurement was less than 5%. Field tests were conducted on a Zoomlion 4YZL-5BZH combined corn kernel harvester and the results showed that the average error of measured yield was 3.72%. Compared to the yield measurement method using the weighing method, the average error of the bench test yield measurement was 7.6% and the average error of yield measurement in field trials with a mass flow sensor yield measurement system was 16.38%. It was verified that the system designed in this study has high yield measurement accuracy and real-time yield measurement, and can provide reference for precision agriculture and high yield management.

The tilting of the cab seat when the tractor is in deep ploughing operation changes the sitting position of the driver, which may accelerate lumbar spine injury. This paper adopts the musculoskeletal model and the finite element model of the lumbar L4–L5 segment to predict the maximum Von-Mises stress and maximum strain of the driver’s lumbar L4–L5 segment intervertebral disc. In this study, we used 3D motion capture to obtain the driver's spine position spatial data when the tractor tilted at different angles. A tractor-driver musculoskeletal model and a finite element model of the lumbar spine L4–L5 segments were created in AnyBody™ and Abaqus, respectively. The tractor-driver musculoskeletal model was used to calculate the load of the driver's lumbar spine L4–L5 segment at different angles of tractor tilt, which was used as the load condition of the finite element model of the lumbar spine L4–L5 segment, and then the influence of tractor tilt angle and vibration on the driver's lumbar spine L4–L5 disc was studied. The results show that the maximum Von-Mises stress and maximum strain of the driver’s lumbar L4–L5 intervertebral disc will increase due to the tilt. The maximum Von-Mises stress occurs in the annulus II, and the maximum strain occurs in the upper end plate of the intervertebral disc. With the occurrence of tilt, the position of the maximum Von-Mises stress changes, which can lead to disc injury to the driver, and vibration may exacerbate this injury.