基于红外热成像的夜间农田实时语义分割

农田环境实时语义分割是构成智能农机的视觉环境感知的重要环节，夜间农田语义分割可以使智能农机在夜间通过视觉感知农田环境进行全天候作业，而夜间无光环境下，可见光摄像头成像效果较差，将造成语义分割精度的下降。为保证夜间农田环境下红外图像语义分割的精度与实时性，该研究提出了一种适用于红外图像的红外实时双边语义分割网络（Infrared Real-time Bilateral Semantic Segmentation Network，IR-BiSeNet），根据红外图像分辨率低，细节模糊的特点该网络在实时双边语义分割网络（Bilateral Semantic Segmentation Net，BiSeNet）结构基础上进行改进，在其空间路径上，进一步融合红外图像低层特征，在该网络构架中的注意力提升模块、特征融合模块上使用全局最大池化层替换全局平均池化层以保留红外图像纹理细节信息。为验证提出方法的有效性，通过在夜间使用红外热成像采集的农田数据集上进行试验，数据集分割目标包括田地、行人、植物、障碍物、背景。经试验验证，提出方法在夜间农田红外数据集上达到了85.1%的平均交并比（Mean Intersection over Union，MIoU），同时达到40帧/s的处理速度，满足对夜间农田的实时语义分割。     

冬小麦品种干旱胁迫后的恢复能力与水分利用效率的关系

为探讨营养生长期干旱胁迫-复水后的恢复力作为品种群体水平水分利用效率(WUE)鉴定指标的可能性,以冬小麦品种‘晋麦47’和‘京411’及其34个近等基因系为材料,利用半自动防雨棚与渗漏池开展模拟干旱实验,于拔节初期干旱较严重时复水,测定复水后的生物量积累作为恢复力,同时测定营养生长期末的株高以及品种/系的群体WUE,分析材料之间恢复力、株高、和群体WUE之间的差异,以及恢复力与群体WUE之间的相关性。结果发现,部分冬小麦品种/系之间恢复力达到显著差异,恢复力受基因型和生长环境共同影响,并且两者存在一定程度的互作。34个近等基因系与亲本在株高和群体WUE上均存在显著差异,品种/系的恢复力与株高之间呈显著的正相关关系,恢复力与株高分别与群体的产量水平的WUE呈极显著正相关关系。该研究结果初步表明,胁迫-复水后的恢复力可以作为品种在干旱条件下群体WUE鉴定评价的指标。     

不同合成树脂工艺对改性材性能影响研究

为研究树脂对改性材性能的影响，采用2种不同工艺合成三聚氰胺-脲醛树脂（MUF），测试了树脂的相关性能。结果表明，不同合成工艺路线下制备的 MUF 树脂在固体含量、粘度、固化时间、游离甲醛含量间存在显著差异。最终树脂的分子结构类型相似性极高，但相同结构组分在不同树脂中所占比例各有差异。羟甲基基团在MUF2中所占比例高，而亚甲基桥键及醚键在MUF1中含量高。MUF1改性材的增重率（weight percent gain，WPG）值更大，但MUF2改性材的抗溶胀性（anti-swelling efficiency，ASE）和体积膨胀率（bulking rates，B）更高，MUF2改性材的尺寸稳定性更好。     

Estimation of litchi (Litchi chinensis Sonn.) leaf nitrogen content at different growth stages using canopy reflectance spectra

The estimation of crop nitrogen status in fresh vegetation leaf using field spectroscopy is challenging due to the weak responses on leaf/canopy reflectance and the overlapping with the absorption features of other compounds. Although the spectral indices were proposed in the literature to predict leaf nitrogen content (LNC), the performance of selected spectral indices to estimate the LNC is often inconsistent. Moreover, the models for nitrogen content estimation changed with the growth stage. The goal of this study was to evaluate the performance of published indices, ratio of data difference index (RDDI) and ratio of data index (RDI) developed by band iterative-optimization algorithm in LNC estimation. The correlation analysis, linear regression and cross validation were used to analyze the relationship between spectral data and LNC and construct the best performed estimation model. The study was conducted by the data of five growing seasons of litchi from the orchards in Guangdong Province of China. Results showed that the relationship between chlorophyll (Chl) related spectral indices and LNC varied with the growth stage. Even in flower bud morphological differentiation stage and autumn shoot maturation stage, there were not significant correlations between the proposed spectral indices and LNC. Besides it is difficult to estimate the LNC by the general model across the growth stages due to the integrated effects of cultivar, biochemical, canopy structure, etc. The band iterative-optimization algorithm can improve the sensitivity of spectral data to LNC to some extent. The optimal RDDI performed better than other indices for the synthetic dataset and the dataset in each growth stage. And the sensitive bands selected in the optimal indices at each growth stage are not consistent, which are not only related to the Chl absorption but also other biochemical components, such as starch, lignin, cellulose, protein, etc. In general, the LNC can be estimated by the optimized CR-based RDDI indices in autumn shoot maturation stage, flower spike stage, fruit maturation stage, and flowering stage with the R² > 0.50 and RMSE < 0.14. Although there were the significant relationship between RDIs and RDDIs in flower bud morphological differentiation stage, the highest R²s of the model developed by RDDIs and RDIs were less than 0.50 in cross validation. This study indicated that the applicability of canopy reflectance to estimate litchi LNC was closely related to the growth stage of litchi. Growth stage-specific models will be preferred for estimating litchi LNC estimation.     

Influence of Soil Particle Size on the Measurement of Sodium by Near-Infrared Reflectance Spectroscopy

This study evaluates the effect of soil particle size (SPS) on the measurement of exchangeable sodium (Na) (EXC-Na) by near-infrared reflectance (NIR) spectroscopy. Three hundred thirty-two (n = 332) top soil samples (0–10 cm) were taken from different locations across Uruguay, analyzed by EXC-Na using emission spectrometry, and scanned in reflectance using a NIR spectrophotometer (1100–2500 nm). Partial least squares (PLS) and principal component regression (PCR) models between reference chemical data and NIR data were developed using cross validation (leaving one out). The coefficient of determination in calibration (R²) and the root mean square of the standard error of cross validation (RMSECV) for EXC-Na concentration were 0.44 (RMSECV: 0.12 mg kg^–1) for soil with small particle size (SPS-0.053) and 0.77 (RMSECV: 0.09 mg kg^–1) for soils with particle sizes greater than 0.212 mm (SPS-0.212), using the NIR region after second derivative as mathematical transformation. The R² and RMSECV for EXC-Na concentration using PCR were 0.54 (RMSECV: 0.07 mg kg^–1) and 0.80 (RMSECV: 0.03 mg kg^–1) for SPS-0.053 and SPS-0.212 samples, respectively.     

利用近红外光谱技术快速分析全株玉米青贮营养成分

研究旨在利用近红外光谱技术(near infrared reflectance spectroscopy,NIRS)建立全株玉米青贮6种营养成分的近红外预测模型,为生产实践中合理利用全株玉米青贮饲料资源提供理论依据。选取玉米青贮样品64份作为定标集,16份作为验证集。利用NIRS结合改良偏最小二乘法(modified partial least squares,MPLS)构建并且验证其建立预测模型的优劣。结果表明:构建的玉米青贮干物质(DM)、粗蛋白(CP)、酸性洗涤纤维(ADF)和粗灰分(Ash)预测模型的交叉验证相关系数(1-VR)分别为0.83、0.80、0.93和0.78,都在0.80左右,可以用于实际预测。粗脂肪(EE)和中性洗涤纤维(NDF)含量预测模型1-VR为0.64和0.40,低于0.80,构建的模型相关性较差,模型还需要进一步优化。     

基于近红外光谱法的紫花苜蓿品质综合评价

探索近红外光谱法测定紫花苜蓿品质的可行性,采用近红外光谱法与常规方法对8个紫花苜蓿品种(系)的粗蛋白、粗脂肪、粗灰分、酸性洗涤纤维、中性洗涤纤维、钙、磷和干物质等8项指标进行测定。利用配对T检验、多重比较和回归分析等方法检验近红外光谱法测定结果的可行性,运用灰色关联度分析对苜蓿品质进行综合评价。结果表明,粗蛋白、粗灰分、中性洗涤纤维、酸性洗涤纤维、磷和干物质等6项指标用近红外光谱法测定可信度较高。紫花苜蓿品质评价指标可简化为粗蛋白、粗灰分、中性洗涤纤维、酸性洗涤纤维和干物质等5项。灰色关联度分析综合表现较好的苜蓿品种(系)有WL358、WL440HQ、WL656HQ和WL366HQ。     

基于梨枣轻微损伤的可见/近红外光谱判别研究

利用可见/近红外光谱技术对梨枣轻微损伤的分类判别建模方法进行研究。分别采用PLS-LDA(线性)和LS-SVM(非线性)建立判别模型,分析比较不同预处理方式和建模波段对模型精度的影响。结果表明:经9点平滑预处理后的短波近红外(780~1100nm)PLS-LDA模型判别效果最佳,校正集和预测集的正确识别率分别达到97.8%和96.7%。     

干旱胁迫对3种冷季型草坪草光谱反射率及生理特征的影响

为探讨高光谱分析技术在草坪草水分监测上的应用,以草地早熟禾、高羊茅、多年生黑麦草为试验材料,测定了干旱胁迫下3种冷季型草坪草生理指标及光谱反射率。结果表明:随着干旱胁迫程度的加深,供试草坪草叶片相对含水量和叶绿素含量显著下降(P0.05),丙二醛含量、脯氨酸含量及敏感波段的(550和760 nm)的光谱反射率显著上升(P0.05),3种草坪草抗旱性能的强弱顺序为:高羊茅多年生黑麦草草地早熟禾。3种草坪草敏感波段光谱反射率(R_(550 nm),R_(760 nm))及高光谱参数(Rg,R0,GNDVI)与生理指标间呈显著或极显著相关,以R_(550 nm)与生理指标间的相关系数绝对值最大。同时R_(550 nm)构建了供试草坪草生理指标的估测模型,经检验,模型估测值与实测值相关性达显著或极显著水平。     

ELLIPSOMETRIC SPECTROSCOPY OF ELECTROCHROMISM OF POLYANILINE FILMS

The electrochromism of polyaniline films, which is prepared by electropolymer-ization from aniline in acidic aqueous solution onto platinum,is investigated in 330-700nm through in-situ ellipsometric technique. The results indicate that the ellipsometric peak of the reduced state(-0. 2V) of the film appears at 390-400nm and the oxidized state(0. 55V) at 500-530nm. The elec-trochromic reaction from the reduced "yellow" state to the oxidized "green" state is suggested via intermediate state whose ellipsometric peak appears at 466-470nm.