基于YOLOv5s改进模型的小白菜虫害识别方法

小白菜是中国种植面积较广、深受大众喜爱的蔬菜,但真实菜地环境中虫害往往出现在叶片的特定区域,且受环境因素如光照和背景干扰较大,影响对其的智能检测。为提高小白菜虫害的检测效率和准确率,该研究提出一种基于YOLOv5s网络框架改进的YOLOPC小白菜虫害识别模型。首先,引入CBAM（Convolutional Block Attention Module）注意力机制,将其放在CBS（卷积层Convolution+归一化层Batch normalization+激活函数层SILU）的输入端构成CBAM-CBS的结构,动态调整特征图中各个通道和空间位置的权重;使用上采样和1×1卷积操作来调整特征图的尺寸和通道数,实现不同层次特征的融合,增强模型的特征表示能力。同时,改进损失函数,使其更适合边界框回归的准确性需求;利用空洞卷积的优势提高网络的感受野范围,使模型能够更好地理解图像的上下文信息。试验结果表明,与改进前的YOLOv5s模型相比,YOLOPC模型对小白菜小菜蛾和潜叶蝇虫害检测的平均精度（mean Average Precision, mAP）达到91.40%,提高了12.9个百分点;每秒传输帧数（Frame Per Second, FPS）为58.82帧/s,增加了11.2帧/s,增加幅度达23.53个百分点;参数量仅为14.4 MB,降低了25.78个百分点。与经典的目标检测算法SSD、Faster R-CNN、YOLOv3、YOLOv7和YOLOv8相比,YOLOPC模型的平均精度分别高出20.1、24.6、14、13.4和13.3个百分点,此外,其准确率、召回率、帧速率和参数量均展现出显著优势。该模型可为复杂背景下小白菜虫害的快速准确检测提供技术支持。     

基于无人机RGB影像的染条锈冬小麦花青素含量监测

小麦植株感染条锈病后叶片花青素含量会发生明显变化。为了在地块尺度上利用冬小麦花青素值实现条锈病害的直观、快速监测,通过监测叶片花青素含量评估小麦条锈病严重程度,2021年获取感染条锈病的小麦田块的无人机RGB影像和采集地面病害区域的花青素含量数据,利用影像提取采样点感兴趣区的光谱特征参数和基于灰度共生矩阵的纹理特征参数,采用连续投影算法(SPA)结合相关性分析优选特征参数,分别采用单一光谱特征参数和组合参数,结合主成分回归(PCR)、拉索回归(LR)、随机森林回归(RFR)、梯度提升回归(GBR)和误差反向传播神经网络(BPNN)等方法构建了小麦花青素含量估算模型,并利用最优模型反演了田块的花青素含量。结果表明,图像光谱特征结合纹理特征后,花青素估算模型的R²增大,RMSE减小,模型精度显著提升。基于组合特征参数构建的随机森林模型精度最高,验证集R²、RMSE和MAE分别为0.801、0.026、0.021。该模型具有良好的花青素含量估算能力,得到的花青素值分布图与条锈病的空间分布具有一致性,能够定量化、可视化地反映病害严重程度。     

基于线性谱聚类的林地图像中枯死树监测

【目的】将基于线性谱聚类超像素的方法应用在森林病虫害防治领域,可智能监测无人机森林虫害图像中的枯死树,为森林有害生物的监测工作提供技术支撑。【方法】分别以湖北省受松材线虫与辽宁省受红脂大小蠹侵害的松林无人机图像为试验数据,首先使用线性谱聚类超像素分割算法将图像划分为多个超像素;然后基于枯死树木的颜色特征,初步提取可能为枯死树的超像素区域;最后基于枯死树木与其他干扰地物具有不同的纹理特征,计算超像素的区域密度和缝隙量,利用支持向量机对初步提取的超像素进行分类,从而检测出图像中的枯死树。【结果】基于线性谱聚类超像素和支持向量机的枯死树监测方法可有效排除与枯死树木颜色相近的其他干扰地物,较准确地提取出枯死树木。使用该方法与基于植被颜色指数的阈值分割方法、基于简单线性迭代聚类超像素和随机森林的方法,对35幅受灾松林无人机图像进行试验,并选用交并比、虚警率和漏检率3个评价指标对3种方法进行定量对比分析。结果表明,基于线性谱聚类超像素的方法监测出的枯死树区域最精确,其监测结果与人工检测结果的交并比均值大于58%,且虚警率和漏检率均优于另外2种方法。【结论】基于线性谱聚类超像素的枯死树监测方法能实现松林中枯死树的快速、准确检测及定位。     

BRAIN MAGNETIC RESONANCE IMAGING CHARACTERISTICS IN DOGS AND CATS WITH CONGENITAL PORTOSYSTEMIC SHUNTS

Animals with a portosystemic shunt (PSS) often have neurologic abnormalities. Diagnostic imaging, including brain magnetic resonance (MR) imaging, is not performed routinely in these animals. In this study, brain MR images were obtained in 13 dogs and three cats with a PSS, and in 15 dogs and five cats that were neurologically normal and used as controls. All animals with a PSS had widened sulci. In addition, 10 out of 13 dogs with a PSS and one out of three cats with a PSS had hyperintense focal areas in the lentiform nuclei on T1-weighted (T1W) images, which did not enhance after intravenous gadolinium. Following surgical correction of the PSS, MR imaging examinations were repeated in one dog and one cat. The hyperintensity of the lentiform nuclei had decreased. This study indicates that MR imaging findings of widened sulci and hyperintensity of the lentiform nuclei on T1W images may be found in dogs and cats with a PSS.     

AN APPLICATION OF IMAGE PROCESSING TECHNIQUES IN COMPUTED TOMOGRAPHY IMAGE ANALYSIS

An estimate of the thickness of subcutaneous adipose tissue at differing positions around the body was required in a study examining body composition. To eliminate human error associated with the manual placement of markers for measurements and to facilitate the collection of data from a large number of animals and image slices, automation of the process was desirable. The open-source and free image analysis program ImageJ was used. A macro procedure was created that provided the required functionality. The macro performs a number of basic image processing procedures. These include an initial process designed to remove the scanning table from the image and to center the animal in the image. This is followed by placement of a vertical line segment from the mid point of the upper border of the image to the image center. Measurements are made between automatically detected outer and inner boundaries of subcutaneous adipose tissue along this line segment. This process was repeated as the image was rotated (with the line position remaining unchanged) so that measurements around the complete circumference were obtained. Additionally, an image was created showing all detected boundary points so that sites of measurement could be checked by visual inspection. This paper provides a detailed account of this macro, illustrates the effect of its components on the image, and the applicability of standard image processing techniques to veterinary imaging.     

鹿产品检测的研究进展

鹿产品是珍贵的中药材,目前市场上存在着严重的以假充真,以次充好的现象,科研工作者对鹿产品的检测做了大量的研究工作。作者回顾了近年来鹿产品检测方面的研究进展,分别从性状检测、显微检测、理化检测及分子检测等几种主要检测方法进行综述,为鹿产品检测的进一步研究提供了参考依据,同时提出了鹿产品检测中存在的问题及进一步研究方向。     

Magnetic resonance imaging of the normal bovine digit

The purpose of this study was defining the normal structures of the digits and hoof in Holstein dairy cattle using Magnetic Resonance Image (MRI). Transverse, Sagital and Dorsoplantar MRI images of three isolated cattle cadaver digits were obtained using Gyroscan T5-NT a magnet of 0.5 Tesla and T1 Weighted sequence. The MRI images were compared to corresponding frozen cross-sections and dissect specimens of the cadaver digits. Relevant anatomical structures were identified and labeled at each level. The MRI images provided anatomical detail of the digits and hoof in Holstein dairy cattle. Transversal images provided excellent depiction of anatomical structures when compared to corresponding frozen cross-sections. The information presented in this paper would serve as an initial reference to the evaluation of MRI images of the digits and hoof in Holstein dairy cattle, that can be used by radiologist, clinicians, surgeon or for research propose in bovine lameness.     

基于"3S"技术的甘肃子午岭自然保护区土地利用变化研究

以甘肃省子午岭自然保护区为研究对象,运用"3S"技术,对研究区的地形图、不同时相的TM遥感影像进行数据预处理,并分类提取了研究区2个时段的土地利用信息.对分类信息进行统计分析后并与实地调查数据进行比较,结果表明,近20年来甘肃子午岭自然保护区的森林覆盖面积在稳步提高,乔木林面积明显增加.为甘肃子午岭自然保护区工程的建设提供了基础信息,为林区植物种群生长及分布调查研究提供了宏观数据,并且输出了研究区不同时相土地利用专题图.     

《天净沙·秋思》英译本之我见——与黄国文、金春笙先生商榷

《中国翻译》上刊登了黄国文和金春笙先生对《天净沙.秋思》英译本的分析和研究。针对金先生和黄先生相驳的两种观点,文章首先从语言学角度审视了“形式对等”的真正内涵、实用性以及可接受性;然后从文学角度——音律、情感逻辑、意象虚实等来探讨《天净沙.秋思》的Schlepp和翁显良两种英译本孰优孰劣;最后建议运用语言学的分析方法来理性思考,结合文学的印象式、感悟式和经验式的翻译批评可以使得译文更富艺术魅力。     

基于高分辨率卫星图像的深圳特区市政绿地结构特征分析

基于“Quickbird”卫星遥感数据对深圳特区市级公园及市政道路绿地现状进行判读,并结合实地调查数据进行比较分析。结果表明,判读结果与实地调查数据相符程度较高,尤其是面积相对较大的公园绿地;对于以乔木为主的分车道绿地,由于判读为覆盖面积,与调查面积出现一定差异;深圳市市政公园绿地中乔灌草所占比例最大,其次为乔草和草坪;市政道路绿地与公园绿地相比,乔木、乔草、灌草、灌木类型比例均有所上升,草坪面积大幅减少,乔灌草复合结构依然占绝对优势。