Grape Berry Detection and Size Measurement Based on Edge Image Processing and Geometric Morphology

Image completion algorithm of anthurium spathes based on multi-scale feature learning

Machine vision has been used to grade the potted anthurium plant in large-scale production recently. Images are taken to measure the number and size of anthurium spathes. However, due to the limitation of the shooting angle, the occlusion problem reduces the accuracy of measurement. It is necessary to segment the overlapping spathes and repair the incomplete ones. The traditional image completion model has good performance on missing small areas, but it is not satisfactory for missing large areas. In this article, a multi-scale fusion Recurrent Feature Reasoning (RFR) network was proposed to repair the spathe images. Unlike the traditional RFR, a multi-layer component was used in the feature reasoning module. This network can combine multi-scale features to complete the learning task and obtain more details of the spathe, which makes the network more advantageous in image completion when missing large areas of spathes. In this study, a comparison experiment between this network and the widely used image completion network was performed, and the results showed that this network performed well in all types of image completion, especially with large-area incomplete images.

Image-Based High-Throughput Phenotyping in Horticultural Crops

Plant phenotyping is the primary task of any plant breeding program, and accurate measurement of plant traits is essential to select genotypes with better quality, high yield, and climate resilience. The majority of currently used phenotyping techniques are destructive and time-consuming. Recently, the development of various sensors and imaging platforms for rapid and efficient quantitative measurement of plant traits has become the mainstream approach in plant phenotyping studies. Here, we reviewed the trends of image-based high-throughput phenotyping methods applied to horticultural crops. High-throughput phenotyping is carried out using various types of imaging platforms developed for indoor or field conditions. We highlighted the applications of different imaging platforms in the horticulture sector with their advantages and limitations. Furthermore, the principles and applications of commonly used imaging techniques, visible light (RGB) imaging, thermal imaging, chlorophyll fluorescence, hyperspectral imaging, and tomographic imaging for high-throughput plant phenotyping, are discussed. High-throughput phenotyping has been widely used for phenotyping various horticultural traits, which can be morphological, physiological, biochemical, yield, biotic, and abiotic stress responses. Moreover, the ability of high-throughput phenotyping with the help of various optical sensors will lead to the discovery of new phenotypic traits which need to be explored in the future. We summarized the applications of image analysis for the quantitative evaluation of various traits with several examples of horticultural crops in the literature. Finally, we summarized the current trend of high-throughput phenotyping in horticultural crops and highlighted future perspectives.

Estimation of Characteristic Parameters of Grape Clusters Based on Point Cloud Data

The measurement of grapevine phenotypic parameters is crucial to quantify crop traits. However, individual differences in grape bunches pose challenges in accurately measuring their characteristic parameters. Hence, this study explores a method for estimating grape feature parameters based on point cloud information: segment the grape point cloud by filtering and region growing algorithm, and register the complete grape point cloud model by the improved iterative closest point algorithm. After estimating model phenotypic size characteristics, the grape bunch surface was reconstructed using the Poisson algorithm. Through the comparative analysis with the existing four methods (geometric model, 3D convex hull, 3D alpha-shape, and voxel-based), the estimation results of the algorithm proposed in this study are the closest to the measured parameters. Experimental data show that the coefficient of determination (R ²) of the Poisson reconstruction algorithm is 0.9915, which is 0.2306 higher than the coefficient estimated by the existing alpha-shape algorithm (R ² = 0.7609). Therefore, the method proposed in this study provides a strong basis for the quantification of grape traits.

Measuring Shape Parameters of Pearls in Batches Using Machine Vision: A Case Study

To solve the problem of low precision of pearl shape parameters’ measurement caused by the mutual contact of batches of pearls and the error of shape sorting, a method of contacting pearls’ segmentation based on the pit detection was proposed. Multiple pearl images were obtained by backlit imaging, the quality of the pearl images was improved through appropriate preprocessing, and the contacted pearl area was extracted by calculating the area ratio of the connected domains. Then, the contour feature of the contact area was obtained by edge tracking to establish the mathematical model of the angles between the edge contour points. By judging the angle with a threshold of 60° as the candidate concave point, a concave point matching algorithm was introduced to get the true concave point, and the Euclidean distance was adopted as a metric function to achieve the segmentation of the tangent pearls. The pearl shape parameters’ model was established through the pearl contour image information, and the shape classification standard was constructed according to the national standard. Experimental results showed that the proposed method produced a better segmentation performance than the popular watershed algorithm and morphological algorithm. The segmentation accuracy was above 95%, the average loss rate was within 4%, and the sorting accuracy based on the shape information was 94%.

Ultra-Wideband Microwave Imaging System for Root Phenotyping

The roots are a vital organ for plant growth and health. The opaque surrounding environment of the roots and the complicated growth process means that in situ and non-destructive root phenotyping face great challenges, which thus spur great research interests. The existing methods for root phenotyping are either unable to provide high-precision and high accuracy in situ detection, or they change the surrounding root environment and are destructive to root growth and health. Thus,we propose and develop an ultra-wideband microwave scanning method that uses time reversal to achieve in situ root phenotyping nondestructively. To verify the method’s feasibility, we studied an electromagnetic numerical model that simulates the transmission signal of two ultra-wideband microwave antennas. The simulated signal of roots with different shapes shows the proposed system’s capability to measure the root size in the soil. Experimental validations were conducted considering three sets of measurements with different sizes, numbers and locations, and the experimental results indicate that the developed imaging system was able to differentiate root sizes and numbers with high contrast. The reconstruction from both simulations and experimental measurements provided accurate size estimation of the carrots in the soil, which indicates the system’s potential for root imaging.

Deep Learning-Based Object Detection System for Identifying Weeds Using UAS Imagery

Current methods of broadcast herbicide application cause a negative environmental and economic impact. Computer vision methods, specifically those related to object detection, have been reported to aid in site-specific weed management procedures for targeted herbicide application within a field. However, a major challenge to developing a weed detection system is the requirement for a properly annotated database to differentiate between weeds and crops under field conditions. This research involved creating an annotated database of 374 red, green, and blue (RGB) color images organized into monocot and dicot weed classes. The images were acquired from corn and soybean research plots located in north-central Indiana using an unmanned aerial system (UAS) flown at 30 and 10 m heights above ground level (AGL). A total of 25,560 individual weed instances were manually annotated. The annotated database consisted of four different subsets (Training Image Sets 1–4) to train the You Only Look Once version 3 (YOLOv3) deep learning model for five separate experiments. The best results were observed with Training Image Set 4, consisting of images acquired at 10 m AGL. For monocot and dicot weeds, respectively, an average precision (AP) score of 91.48 % and 86.13% was observed at a 25% IoU threshold (AP @ T = 0.25), as well as 63.37% and 45.13% at a 50% IoU threshold (AP @ T = 0.5). This research has demonstrated a need to develop large, annotated weed databases to evaluate deep learning models for weed identification under field conditions. It also affirms the findings of other limited research studies utilizing object detection for weed identification under field conditions.