To clarify the differences in and mechanisms of soil detachment before and after soil collapse, five typical granite soil layers (red soil, red soil-sandy soil, sandy soil, sandy soil-debris, and debris layers) of Benggang in Anxi County, Fujian Province, were studied via laboratory runoff scouring tests, and the detachment capabilities and influencing factors of undisturbed (original) and disturbed (colluvial deposit) soils were compared. The results showed that disturbance due to soil collapse significantly increases the soil detachment capacity by an average of 1046 times, with the greatest increase occurring in the red soil-sand soil layer (3494 times) and the smallest increase occurring in the debris layer (63 times). The undisturbed soil detachment capacity increases with increasing soil depth, whereas the disturbed soil capacity first increases but then decreases, with the sand layer having the highest capacity. Hydrodynamic fitting results revealed that undisturbed red soil has a linear relationship, red soil-sandy soil and sandy soil layers have power function relationships, and sandy soil-debris and debris layers have logarithmic relationships with flow shear stress. Disturbed red soil and red soil-sandy soil layers are linearly related, whereas the other layers are logarithmically related. Correlation analysis revealed that undisturbed soil detachment is significantly negatively correlated with clay, silt, gravel, free iron oxide, and free alumina contents and positively correlated with sand content. Disturbed soil shows similar correlations, but it has a negative correlation with organic matter instead of gravel. Structural equation modelling (SEM) path analysis revealed that undisturbed soil detachment is affected mainly by negative free alumina oxide content (path coefficient of -0.87) and flow shear stress (path coefficient of 0.14), whereas disturbed soil is controlled mainly by negative shear strength (path coefficient of -0.76) and positive flow shear stress (path coefficient of 0.49). This study elucidates the mechanism by which colluvial deposit disturbance accelerates soil detachment, providing a theoretical basis for the prevention and control of Benggang erosion in the hilly regions of southern China with red soil. Moreover, the comparative research strategy adopted in this study offers a reference for related investigations in similar erosion-prone areas.

Traditional image classification often misclassifies unknown samples as known classes during testing, degrading recognition accuracy. Open-set image recognition can simultaneously detect known classes (KCs) and unknown classes (UCs) but still struggles to improve recognition performance caused by open space risk. Therefore, we introduce a cosine distance loss function (CDLoss), which exploits the orthogonality of one-hot encoding vectors to align known samples with their corresponding one-hot encoder directions. This reduces the overlap between the feature spaces of KCs and UCs, mitigating open space risk. CDLoss was incorporated into both Softmax-based and prototype-learning-based frameworks to evaluate its effectiveness. Experimental results show that CDLoss improves AUROC, OSCR, and accuracy across both frameworks and different datasets. Furthermore, various weight combinations of the ARPL and CDLoss were explored, revealing optimal performance with a 1:2 ratio. T-SNE analysis confirms that CDLoss reduces the overlap between the feature spaces of KCs and UCs. These results demonstrate that CDLoss helps mitigate open space risk, enhancing recognition performance in open-set image classification tasks.

Pine wilt disease poses a serious threat to pine forests worldwide. It causes rapid pine tree mortality, profoundly impacting ecosystems and economic assets. Using unmanned aerial vehicle (UAV) imagery to identify, locate, and promptly remove dead pine trees in forest areas is an effective method to control the spread of pine wilt disease. However, dataset annotation and validation remains challenging. To address the high cost and time-consuming nature of manual annotation in fully supervised learning for identifying dead pine trees, this study introduces semi-supervised learning framework for detecting dead pine trees. Unlike traditional fully supervised learning methods, our approach reduces the need for a large quantity of precisely annotated data. By incorporating attention-enhanced pyramid and adaptive thresholding mechanisms, the model can effectively learn from a small amount of labelled data and generate high-quality pseudo-labels from unlabelled data. Specifically, the model employs the channel prior convolutional attention mechanism in feature pyramid fusion to enhance sensitivity to key features of dead trees and improve consistency learning. Meanwhile, probabilistic clustering Gaussian Mixture Model is also used to adaptively adjust the threshold for pseudo-labels assignment, enhancing the accuracy of pseudo-labels during training. Experiments were conducted using self-constructed dead pine tree image datasets to validate the effectiveness of the proposed method. The results show that our method requires only 25% of the labelled data to achieve recognition results close to those of fully supervised method with a fully labelled. Additionally, its performance surpasses existing semi-supervised methods. The proposed algorithm provides an efficient and cost-effective approach for detecting dead pine trees for pine wilt disease control using UAV imagery.

Assessing the distribution and risks associated with the soil lead content in the Tieguanyin tea plantations of Anxi County is critical, given the county’s significance as the primary Tieguanyin tea production area in Fujian Province. This study examined the distribution characteristics of soil lead in Anxi County’s tea plantations according to the Kriging spatial interpolation of the parameters of the semivariance function of the exponential model. Moreover, the sources of lead content were analyzed, considering geological backgrounds and anthropogenic influences. Ecological risks and the issuance of early warnings were also assessed. The soil lead content in the rocks of the Tieguanyin tea plantations in Anxi County followed the order: andesite > dacite > rhyolite > granite. The soil lead content gradually decreased from the center toward the east and west, forming four distinct north–south parallel zones. High-lead-content areas were identified at the border of Jiandou, Bailai, and Hushang; in the central part of Lutian; and in the southern part of Huqiu. The high levels of soil lead in the tea plantations possibly originated from industrial and mining activities, automobile exhaust, and agricultural activities. The distribution of single-factor pollution indices and potential risk evaluation based on the Soil Environmental Quality Standard, Environmental Technical Conditions for Tea Production Area, and Environmental Technical Conditions for Organic Tea Production Area indicated that the soil in Tieguanyin tea plantations in Anxi County was clean and safe for tea cultivation.

The detachment–transport coupling equation by Foster and Meyer is a classical equation that describes the relationship between detachment and transport. The equation quantifies the relationship between sediment loads and soil detachment rates, deepens the understanding of soil erosion and provides a reliable basis for the establishment of an erosion model. However, the applicability of this equation to slopes with gradients greater than 47% is limited. In this work, the detachment–transport coupling relationship is investigated using the colluvium material of Benggang. A nonerodible rill flume 4 m long and 0.12 m wide was adopted. The slope gradient ranged from 27% to 70%, the unit flow discharge ranged from 0.56 × 10−3 to 3.33 × 10−3 m2 s−1, and the sediment transport capacity (Tc) was measured under each slope and discharge combination. The sediment was inputted into the flume according to the predetermined sediment addition rate (from 0% to 100% of Tc), and the detachment rate (Dr) under each combination of the slope and discharge was measured. Dr linearly decreased with increasing sediment loads, which is consistent with the detachment–transport coupling equation by Foster and Meyer. The linear equations can predict the detachment capacity (Dc) and Tc well (Nash–Sutcliffe efficiency coefficient (NSE) = 0.98 for Dc, and NSE = 0.99 for Tc). The detachment–transport coupling equation can adequately predict the Dr (NSE = 0.89). However, its applicability to slopes of <47% (NSE: 0.92–0.96) was greater than that to slopes of ≥47% (NSE: 0.81–0.89), and the predicted Dr under Tc levels of 20% and 40% were higher than the measured values, while the predicted value under a Tc level of 80% was lower than the measured value. In summary, the detachment–transport coupling equation by Foster and Meyer can accurately reflect the negative feedback relationship between detachments and transports along steep-slope fixed beds and is suitable for colluvial deposit research. The results provide a basis for the construction of steep-slope colluvial deposit erosion models. In the future, the study of the hydrodynamic characteristics of sediment transport processes should be strengthened to clarify the detachment–transport effect of flows through hydrodynamics.

Pine wilt disease has posed a significant threat to forest ecosystems,due to its highly contagious and destructive nature.The critical step in the prevention and control of pine wilt is eliminating the disease sources,which requires the accurate recognition and removal of dead pine trees.However,small or blurred targets are captured,such as overexposure,backlight,and samples occluded by foliage in practical applications.The reason is that the UAVs have to fly high for capture,due to the geography of the hilly and mountain areas.In this study,a novel You Look Only Once v5（YOLOv5） algorithm was proposed for the recognition of dead pine trees.The super-resolution reconstruction was performed at the feature level,in order to overcome the challenge of recognizing such targets.The YOLOv5 structure was redesigned in two aspects.Firstly,the Selective Kernel Feature Texture Transfer（SKFTT） module was adopted to create the high-resolution detection feature maps with detailed textures,where improved detection accuracy was obtained for the small targets and blurred targets.Specifically,the feature maps with high texture were selected from the backbone network,whereas,the feature maps with high semantics were selected from the feature fusion network.These feature maps were then sent to the texture extractor and content extractor.A selective feature fusion module was used to fuse the critical information about different scales using their weights.Secondly,the Foreground Background Loss function（FB Loss） was introduced to attenuate useless features,while enhancing the gradient contribution of positive samples,and balancing the distribution of positive and negative samples,in order to supervise the reconstruction of high-resolution feature maps.Furthermore,the dataset was obtained to validate the effectiveness of the improved model from the approximately 15 400 hectares of forest land located in Fuzhou and Minhou City,Fujian Province,China.The UAV images were subsequently cropped and screened to obtain about 29 250 labelled samples for further experiments.A series of ablation tests and visualizations were conducted on the testing datasets to verify the effectiveness.Experimental results showed that the mean Average Precision(mAP 50 ) of the improved model was 92.7%,mAP 50～95 was62.1%,and AP small was 53.2%.Compared with the baseline model,the improved model was achieved in the increases of 3.2,8.3,and 15.8 percentage points in the mAP 50 ,mAP 50～95 ,and mAP small ,respectively.The mAP 50 of the improved model was16.7,15.3,2.5,2.8,12.3,and 1.2 percentage points higher than that of the Faster R-CNN,YOLOv4,YOLOX,MT-YOLOv6,QueryDet,and DDYOLOv5 networks,respectively.In addition,the improved model was achieved in the frames per second EPS of 37,which fully met the detection requirements of dead pine trees.Visualization results showed that the improved model can be expected to serve as the recognition of occlusion,overexposure,and backlight targets.The feature maps of the small target detection layer were visualized with different super-resolution algorithms to facilitate observation.The comparison revealed that the texture was improved with the apparently clear boundary shape.In conclusion,the detecting challenge of small targets and blurred targets can be effectively alleviated using the improved dead pine tree detection algorithm,due to the high accuracy.Therefore,the improved detection algorithm is conducive to the efficient removal and comprehensive prevention/control of diseased trees.The improved model can greatly contribute to accelerating the "digital forest prevention" pro cess in precision agriculture.

Accurately segmenting an insect from its original ecological image is the core technology restricting the accuracy and efficiency of automatic recognition. However, the performance of existing segmentation methods is unsatisfactory in insect images shot in wild backgrounds on account of challenges: various sizes, similar colors or textures to the surroundings, transparent body parts and vague outlines. These challenges of image segmentation are accentuated when dealing with camouflaged insects. Here, we developed an insect image segmentation method based on deep learning termed the progressive refinement network (PRNet), especially for camouflaged insects. Unlike existing insect segmentation methods, PRNet captures the possible scale and location of insects by extracting the contextual information of the image, and fuses comprehensive features to suppress distractors, thereby clearly segmenting insect outlines. Experimental results based on 1900 camouflaged insect images demonstrated that PRNet could effectively segment the camouflaged insects and achieved superior detection performance, with a mean absolute error of 3.2%, pixel-matching degree of 89.7%, structural similarity of 83.6%, and precision and recall error of 72%, which achieved improvements of 8.1%, 25.9%, 19.5%, and 35.8%, respectively, when compared to the recent salient object detection methods. As a foundational technology for insect detection, PRNet provides new opportunities for understanding insect camouflage, and also has the potential to lead to a step progress in the accuracy of the intelligent identification of general insects, and even being an ultimate insect detector.

Salient object detection (SOD) aims at highlighting important foreground objects automatically from the background. Most existing SOD methods only employ visible images (RGB images) for salient detection, which limits the performance of real-life applications when encountering challenging scenarios such as low illumination, haze, and smog. In this paper, we take advantage of the RGB and thermal images and propose an Enhancement and Aggregation–Feedback Network (EAF-Net) for SOD. Specifically, to achieve effective complementation between modalities and prevent the interference from noises, we first treat RGB and thermal images equally in the Feature Enhancement Block (FEB), and further, the Global Context Module expands receptive field to obtain the global features and the Top-Feature Enhancement Module suppresses the redundant information that may destroy the original features from the top layer. Subsequently, we embed several Cross Feature Aggregation Modules (CFAMs) into the Aggregation-and-Feedback Decoder to fuse different level features and compensation features for further obtaining comprehensive feature expression. Moreover, a feedback mechanism is adopted to propagate these fused features back into previous layers for refinement and generate saliency maps to decode features in a progressive way. Comprehensive experiments on RGB-T datasets demonstrate that EAF-Net achieves outstanding performance against the state-of-the-art models.

Failure of collapsing walls is an important process affecting the development of Benggang and is closely related to the soil shear strength. Plant roots can increase the soil shear strength. However, the effects and mechanisms of root reinforcement on the soil shear strength of collapsing walls remain unclear. To explore the shear strength characteristics of collapsing walls and their influencing factors under different vegetation conditions, Pennisetum sinese, Dicranopteris dichotoma, Odontosoria chinensis, and Neyraudia reynaudiana were adopted as experimental objects in the Benggang district of Anxi County, Southeast China. We measured the root characteristics and in situ shear strength of root–soil complexes by dividing soil with the four vegetation conditions into five soil layers: 0–5 cm, 5–10 cm, 10–15 cm, 15–20 cm, and 20–25 cm. The average shear strength of the root–soil complexes of the various plants ranked as follows: Pennisetum sinese (30.95 kPa) > Odontosoria chinensis (28.08 kPa) > Dicranopteris dichotoma (21.24 kPa) > Neyraudia reynaudiana (14.99 kPa) > bare soil (11.93 kPa). The enhancement effect of the root system on the soil shear strength was mainly manifested in the 0–5 cm soil surface layer. The soil shear strength attained an extremely significant positive correlation with the root length density, root surface area density, root volume density, root biomass density, for root diameters (L) less than or equal to 0.5 mm and between 0.5 and 1 mm, the soil shear strength could be simulated by using root volume density. The shear strength of undisturbed root–soil complexes measured with a 14.10 pocket vane tester was higher than the value obtained with the Wu–Waldron model (WWM). The correction coefficient k′ varied between 0.20 and 20.25, mostly exceeding 1, and the average correction coefficient k′ value was 4.94. The average correction coefficient determined in this test can be considered to modify the WWM model when conducting experiments under similar conditions.

Low-light image enhancement is a low-level task that aims to improve the brightness and contrast of underexposed images. State-of-the-art low-light methods for image enhancement, benefiting from the development of deep learning, have made great progress. The majority of existing approaches, however, are based on convolutional neural networks, and just a few attempts have been made with Transformers which could perform admirably on high-level vision tasks by readily building long-range dependencies and global context connections. In this paper, we propose LLU-Swin, a hierarchical encoder-decoder architecture, based on Swin Transformer and U-Net for low-light image enhancement. LLU-Swin is composed of several Residual Recovery Transformer Modules (RRTM), each of which contains several improved Swin Transformer layers with a residual connection. We also show Dilated Local-enhanced Window Transformer Block (DLTB), which uses non-overlapping window-based self-attention to offer tremendous efficiency and employs Dilated Locally-enhanced Feed-Forward Network(D-LeFF) to enhance the local sensory ability. Benefits from these two designs, LLU-Swin has great power to capture both global and local contexts. Experimental results demonstrate the LLU-Swin outperforms the state-of-the-art methods in image quality and inference speed.

针对帝王蝶优化算法用于特征选择时需满足多目标的要求,对该算法进行了3个方面的改进:1)在个体排序步骤中引入非支配排序算法,并对调整算子做了修正,满足了多目标要求;2)增加了准确度优先策略,减少了计算资源在低准确性区域的搜索,保证了模型的准确性,满足了特征选择中准确性优先于特征数的要求;3)增加了基于子组的突变策略,对不同子组使用不同的突变策略,避免了算法过早陷入局部最优,解决了算法早熟问题.在3个定量构效特征选择基准数据集上进行了一系列实验,实验结果表明改进的算法与其它算法相比显著提高了模型的准确性并减少了特征数,证明了改进策略的有效性.

In this paper, we conduct a comprehensive study on insect identification and classification. However, the existing insect datasets are made up of several categories, which is far from the demands in reality. To handle this issue, we contribute to a more challenging insect image dataset, which contains 1848 images covering different pests from 118 classes. We further conduct fine-tuning on seven deep convolution neural networks, including VGG16, ResNet50, DenseNet121, Res2Net50_26w_ 4s, SCNet50_ vld, GhostNet, and RegNet. Finally, extensive experiments of the aforementioned networks on our dataset illustrate that the last four state-of-the-art deep convolution neural networks can achieve promising performance on pest identification and classification.

[目的]具有复杂背景的蝴蝶图像前背景分割难度大.本研究旨在探索基于深度学习显著性目标检测的蝴蝶图像自动分割方法.[方法]应用DUTS-TR数据集训练F3Net显著性目标检测算法构建前背景预测模型,然后将模型用于具有复杂背景的蝴蝶图像数据集实现蝴蝶前背景自动分割.在此基础上,采用迁移学习方法,保持ResNet骨架不变,利用蝴蝶图像及其前景蒙板数据,使用交叉特征模块、级联反馈解码器和像素感知损失方法重新训练优化模型参数,得到更优的自动分割模型.同时,将其他5种基于深度学习显著性检测算法也用于自动分割,并比较了这些算法和F3Net算法的性能.[结果]所有算法均获得了很好的蝴蝶图像前背景分割效果,其中,F3Net是更优的算法,其7个指标S测度、E测度、F测度、平均绝对误差(MAE)、精度、召回率和平均IoU值分别为0.940,0.945,0.938,0.024,0.929,0.978和0.909.迁移学习则进一步提升了F3Net的上述指标值,分别为0.961,0.964,0.963,0.013,0.965,0.967和0.938.[结论]研究结果证明结合迁移学习的F3Net算法是其中最优的分割方法.本研究提出的方法可用于野外调查中拍摄的昆虫图像的自动分割,并拓展了显著性目标检测方法的应用范围.

利用诱捕技术监测和控制松墨天牛Monochamus alternatus已成为松材线虫病防控的一项重要措施.利用ionoic、Spring Boot和MyBatis框架,对如何构建松墨天牛诱捕器的智能管理系统进行了研究.结果表明:开发的系统通过浏览器端主页给用户提供的功能包括生成和管理二维码、查看和汇总诱捕数据、诱捕器空间分布可视化展示和统计分析等,通过APP端提供给用户的功能包括扫描二维码获取诱捕器空间位置信息、自动解析行政区划、录入诱捕情况、不同网络状态下的数据远程传输等.APP使用简便,作业人员在几分钟内即可学会熟练使用.系统已部署到云服务器上,并管理近4万个诱捕器,运行情况正常.平台在福建和山东省的成功应用表明,本系统可以实现对大量诱捕器的高效数据采集、管理和统计学分析,解决了人工记录和管理野外诱捕信息费时费力易混乱的问题.

Most of insects of moth species are the agricultural pests,therefore the automatic recognition of the category of insects of moth species are of great importance for pest prediction.Current automatic insect recognition methods mostly focus on recognising the insects above the order level but are hard to recognise the insects on the order level,in light of this problem we proposed a texture feature-based recognition method for insects of moth species.It extracts the features of insect images by applying a modified local binary pattern algorithm (CLBP),and distils nature dimension from the extracted feature matrix according to the category of insects of moth species,and finally applies KNN algo-rithm to the realisation of recognising insects of moth species.Experimental results showed that the discriminative CLBP can extract features from original insect images directly and achieves excellent recognition performance,meanwhile it can effectively reduce the dimensions of fea-tures matrix,so that dwindles the storage space and decreases the computational complexity of similarity comparison.Our research broadens the application scope of computer vision and will be helpful for early prediction of agriculture pests and diseases.

To alleviate identification problems attributed to divergent structure and low conservative of piRNA, combined method based on support vector machine ( SVM) and K-mer was applied to predict sequence of piRNA. To start with, piRNA sequences from non-coding RNA database of several species were identified as positive sample, and from the same database non-piRNA se-quences were set as negative sample. Positive and negative samples were reconstituted as a new dataset, half of which were random-ized as training dataset and the remaining as testing dataset. Subsequently, feature matrix was derived from K-mer distribution which was based on positive and negative sequences, and followed by being classified by SVM and piRNA prediction. Results revealed that K-mer-SVM had higher accuracy, sensitivity and specificity, and better classifying performance of MCC and F-meature than those of K-mer-LDA. In other word, K-mer-SVM classifier is likely to be a better algorithm for piRNA prediction.

Histogram-based methods are widely used in computer vision and there are many histogram comparison algorithms to increase the performance of information retrieval.However,histogram comparison algorithms for 3D model retrieval are seldom studied.Therefore,we use histogram of oriented gradients to extract features of 3D models and then propose a histogram comparison algorithm for 3D model retrieval.The experimental results show that the histogram comparison algorithm does improve the retrieval performance of some classes of models from Princeton Shape Benchmark.

In the research,the principle of color constancy is introduced to differentiate differential structure in nature images and distance regularized level set method is combined in order to segment images with shadows and/or highlights.Experimental results show that the new algorithm can segment the target object effectively and precisely,however,according to the same images the algorithm without introduction of color constancy get wrong segmentation results.

Fast level set method has been applied to grey images segmentation. In order to implement segmentation for images with shadows and/or highlights, the dichromatic reflection model is combined with fast level set method. Experimental results show that fast level set method with the dichromatic reflection model can segment the image with shadows and highlights successfully, while the method without the support of dichromatic reflection model can not. The performance of the new method is better than fast level set method without the support of dichromatic reflection model.

Edge flow-based image segmentation algorithm is to detect edges of object through the location where the edge flows point each other,which solve the problem of determine suitable threshold in traditional edge-based image segmentation algorithms.Therefore,the principle of edge flow based image segmentation algorithm is introduced and the algorithm is modified for image segmentation of insects.The experimental results shown that the algorithm not only segment whole insect object effectively but also segment different parts of one insect.Comparing traditional edge based image segmentation algorithm the algorithm is better.research accelates the applitheion vision technology to agriculture and forest field.