RESUMEN
Peanut pod rot is one of the major plant diseases affecting peanut production and quality over China, which causes large productivity losses and is challenging to control. To improve the disease resistance of peanuts, breeding is one significant strategy. Crucial preventative and management measures include grading peanut pod rot and screening high-contributed genes that are highly resistant to pod rot should be carried out. A machine vision-based grading approach for individual cases of peanut pod rot was proposed in this study, which avoids time-consuming, labor-intensive, and inaccurate manual categorization and provides dependable technical assistance for breeding studies and peanut pod rot resistance. The Shuffle Attention module has been added to the YOLOv5s (You Only Look Once version 5 small) feature extraction backbone network to overcome occlusion, overlap, and adhesions in complex backgrounds. Additionally, to reduce missing and false identification of peanut pods, the loss function CIoU (Complete Intersection over Union) was replaced with EIoU (Enhanced Intersection over Union). The recognition results can be further improved by introducing grade classification module, which can read the information from the identified RGB images and output data like numbers of non-rotted and rotten peanut pods, the rotten pod rate, and the pod rot grade. The Precision value of the improved YOLOv5s reached 93.8%, which was 7.8%, 8.4%, and 7.3% higher than YOLOv5s, YOLOv8n, and YOLOv8s, respectively; the mAP (mean Average Precision) value was 92.4%, which increased by 6.7%, 7.7%, and 6.5%, respectively. Improved YOLOv5s has an average improvement of 6.26% over YOLOv5s in terms of recognition accuracy: that was 95.7% for non-rotted peanut pods and 90.8% for rotten peanut pods. This article presented a machine vision- based grade classification method for peanut pod rot, which offered technological guidance for selecting high-quality cultivars with high resistance to pod rot in peanut.
RESUMEN
Gossypol, as an important oil and raw material for feed, is mainly produced by cotton pigment gland, and has a wide range of applications in the fields of pharmaceutics, agriculture and industry. Accurate knowledge of the distribution of pigment gland in cotton leaves is important for estimating gossypol content. However, pigment glands are extremely small and densely distributed, manual counting is laborious and time-consuming, and difficult to count quickly and accurately. It is thus necessary to design a fast and accurate gland counting method. In this paper, the machine vision imaging technology is used to establish an image acquisition platform to obtain cotton leaf images, and a network structure is proposed based on deep learning, named as Interpolation-pooling net, to segment the pigment glands in the cotton leaf images. The network adopts the structure of first interpolation and then pooling, which is more conducive to the extraction of pigment gland features. The accuracy of segmentation of the model in cotton leaf image set is 96.7%, and the mIoU (Mean Intersection over Union), Recall, Precision and F1-score is 0.8181, 0.8004, 0.8004 and 0.8004 respectively. In addition, the number of pigment glands in cotton leaves of three different densities was measured. Compared with manual measurements, the square of the correlation coefficient (R 2) of the three density pigment glands reached 0.966, 0.942 and 0.91, respectively. The results show that the proposed semantic segmentation network based on deep learning has good performance in the detection and counting of cotton pigment glands, and has important value for evaluating the gossypol content of different cotton varieties. Compared with the traditional chemical reagent determination method, this method is safer and more economical.
RESUMEN
The Rhizotrons method is an important means of detecting dynamic growth and development phenotypes of plant roots. However, the segmentation of root images is a critical obstacle restricting further development of this method. At present, researchers mostly use direct manual drawings or software-assisted manual drawings to segment root systems for analysis. Root systems can be segmented from root images obtained by the Rhizotrons method, and then, root system lengths and diameters can be obtained with software. This type of image segmentation method is extremely inefficient and very prone to human error. Here, we investigate the effectiveness of an automated image segmentation method based on the DeepLabv3+ convolutional neural network (CNN) architecture to streamline such measurements. We have improved the upsampling portion of the DeepLabv3+ network and validated it using in situ images of cotton roots obtained with a micro root window root system monitoring system. Segmentation performance of the proposed method utilizing WinRHIZO Tron MF analysis was assessed using these images. After 80 epochs of training, the final verification set F1-score, recall, and precision were 0.9773, 0.9847, and 0.9702, respectively. The Spearman rank correlation between the manually obtained Rhizotrons manual segmentation root length and automated root length was 0.9667 (p < 10-8), with r 2 = 0.9449. Based on the comparison of our segmentation results with those of traditional manual and U-net segmentation methods, this novel method can more accurately segment root systems in complex soil environments. Thus, using the improved DeepLabv3+ to segment root systems based on micro-root images is an effective method for accurately and quickly segmenting root systems in a homogeneous soil environment and has clear advantages over traditional manual segmentation.
RESUMEN
The number of chemical species is crucial in analyzing pulsed field gradient nuclear magnetic resonance spectral data. Any method to determine the number must handle the obstacles of collinearity and noise. Collinearity in pulsed field gradient NMR data poses a serious challenge to and fails many existing methods. A novel method is proposed by taking advantage of the two obstacles instead of eliminating them. In the proposed method, the determination is based on discriminating decay-profile-dominant eigenvectors from noise-dominant ones, and the discrimination is implemented with a novel low- and high-frequency energy ratio (LHFER). Its performance is validated with both simulated and experimental data. The method is mathematically rigorous, computationally efficient, and readily automated. It also has the potential to be applied to other types of data in which collinearity is fairly severe.
RESUMEN
A large-scale bioreactor experiment lasting for 2years was presented in this paper to investigate the biochemical, hydrological and mechanical behaviors of high food waste content (HFWC) MSW. The experimental cell was 5m in length, 5m in width and 7.5m in depth, filled with unprocessed HFWC-MSWs of 91.3 tons. In the experiment, a surcharge loading of 33.4kPa was applied on waste surface, mature leachate refilling and warm leachate recirculation were performed to improve the degradation process. In this paper, the measurements of leachate quantity, leachate level, leachate biochemistry, gas composition, waste temperature, earth pressure and waste settlement were presented, and the following observations were made: (1) 26.8m3 leachate collected from the 91.3 tons HFWC-MSW within the first two months, being 96% of the total amount collected in one year. (2) The leachate level was 88% of the waste thickness after waste filling in a close system, and reached to over 100% after a surcharge loading of 33.4kPa. (3) The self-weight effective stress of waste was observed to be close to zero under the condition of high leachate mound. Leachate drawdown led to a gain of self-weight effective stress. (4) A rapid development of waste settlement took place within the first two months, with compression strains of 0.38-0.47, being over 95% of the strain recorded in one year. The compression strain tended to increase linearly with an increase of leachate draining rate during that two months.
Asunto(s)
Alimentos , Eliminación de Residuos/métodos , Instalaciones de Eliminación de Residuos , Residuos/análisis , Biodegradación Ambiental , Reactores Biológicos , HidrologíaRESUMEN
2D diffusion-ordered NMR spectroscopy (DOSY) has been widely recognized as a powerful tool for analyzing mixtures and probing inter-molecular interactions in situ. But it is difficult to differentiate molecules with similar diffusion coefficients in presence of overlapped spectra. Its performance is susceptible to the number of chemical components, and usually gets worse when the number of components increases. Here, to alleviate the problem, numerical simultaneous inversion of Laplace transform (SILT) of many related variables is proposed for reconstructing DOSY spectrum (SILT-DOSY). The advantage of the proposed method in comparison to other methods is that it is capable of estimating the number of analytes more accurately and deriving corresponding component spectra, which in turn leads to the more reliable identification of the components.
RESUMEN
Phosphorus accumulating organisms (PAOs) have been found to act as glycogen-accumulating organisms (GAOs) under certain conditions, thus, the deterioration in the performance of enhanced biological phosphorus removal systems is not always attributed to the proliferation of GAOs. In this work, the effects of calcium on the metabolic pathway of PAOs were explored. It was found that when the influent Ca(2+) concentration was elevated, the tendency and extent of extracellular calcium phosphate precipitation increased, and the intracellular inert Ca-bound polyphosphate was synthesized, while the microbial population remained almost unchanged. The changes in the ratios of phosphorus released/acetate uptaken, the glycogen degraded/acetate uptaken and the poly-ß-hydroxyalkanoates synthesized/acetate uptaken during the anaerobic period confirm that, as the influent Ca(2+) concentration was increased, the polyphosphate-accumulating metabolism was partially shifted to the glycogen-accumulating metabolism. At an influent Ca(2+) around 50 mg/L, in addition to the extracellular calcium phosphate precipitation, the intracellular inert Ca-bound polyphosphate synthesis might also be involved in the metabolic change of PAOs. The results of the present work would be beneficial to better understand the biochemical metabolism of PAOs in enhanced biological phosphorus removal systems.
Asunto(s)
Reactores Biológicos , Calcio/metabolismo , Fósforo/metabolismo , Glucógeno/metabolismo , Aguas del Alcantarillado/microbiología , Purificación del AguaRESUMEN
Donkey meat samples (n = 167) from different parts of donkey body (neck, costalia, rump, and tendon), beef (n = 47), pork (n = 51) and mutton (n = 32) samples were used to establish near-infrared reflectance spectroscopy (NIR) classification models in the spectra range of 4,000~12,500 cm(-1). The accuracies of classification models constructed by Mahalanobis distances analysis, soft independent modeling of class analogy (SIMCA) and least squares-support vector machine (LS-SVM), respectively combined with pretreatment of Savitzky-Golay smooth (5, 15 and 25 points) and derivative (first and second), multiplicative scatter correction and standard normal variate, were compared. The optimal models for intact samples were obtained by Mahalanobis distances analysis with the first 11 principal components (PCs) from original spectra as inputs and by LS-SVM with the first 6 PCs as inputs, and correctly classified 100% of calibration set and 98. 96% of prediction set. For minced samples of 7 mm diameter the optimal result was attained by LS-SVM with the first 5 PCs from original spectra as inputs, which gained an accuracy of 100% for calibration and 97.53% for prediction. For minced diameter of 5 mm SIMCA model with the first 8 PCs from original spectra as inputs correctly classified 100% of calibration and prediction. And for minced diameter of 3 mm Mahalanobis distances analysis and SIMCA models both achieved 100% accuracy for calibration and prediction respectively with the first 7 and 9 PCs from original spectra as inputs. And in these models, donkey meat samples were all correctly classified with 100% either in calibration or prediction. The results show that it is feasible that NIR with chemometrics methods is used to discriminate donkey meat from the else meat.
Asunto(s)
Equidae , Carne/clasificación , Espectroscopía Infrarroja Corta , Animales , Calibración , Bovinos , Análisis de los Mínimos Cuadrados , Modelos Teóricos , Máquina de Vectores de Soporte , PorcinosRESUMEN
The choice of the type of background spectrum affects the credibility of open-path Fourier transform infrared spectroscopy (OP/FT-IR) data, and consequently, the quality of data analysis. We systematically investigated several properties of the background spectrum. The results show that a short-path background measured with the lowest amplifier gain could significantly reduce noise in the calculated absorbance spectrum, by at least 30% in our case. We demonstrated that by using a short-path background, data analysis is more resistant to interferences such as wavenumber shift or resolution alteration that occurs as a consequence of aging hardware or misalignment. We discussed a systematic error introduced into quantitative analyses by the short-path background and developed a procedure to correct that error. With this correction approach, a short-path background established five years ago was still found to be valid. By incorporating these findings into the protocol for quantitative analysis, we processed the measurements with two OP/FT-IR instruments set up side by side in the vicinity of a large dairy farm, to monitor NH3, CH4, and N2O. The two sets of calculated concentrations showed high agreement with each other. The findings of our investigations are helpful to atmospheric monitoring practitioners of OP/FT-IR spectroscopy and could also be a reference for future amendments to the protocols outlined in the guidelines of the U.S. Environmental Protection Agency, the American Society for Testing and Materials, and the European Committee for Standardization.
RESUMEN
Strawberry variety is a main factor that can influence strawberry fruit quality. The use of near-infrared reflectance spectroscopy was explored discriminate among samples of strawberry of different varieties. And the significance of difference among different varieties was analyzed by comparison of the chemical composition of the different varieties samples. The performance of models established using back propagation-artificial neural networks (BP-ANN), least squares-support vector machine and discriminant analysis were evaluated on spectra range of 4545-9090 cm(-1). The optimal model was obtained by BP-ANN with a topology of 12-18-3, which correctly classified 96.68% of calibration set and 97.14% of prediction set. And the 94.95%, 97% and 98.29% classifications were given respectively for "Tianbao" (n=99), "Fengxiang" (n=100) and "Mingxing" (n=117). One-way analysis of variance was made for comparison of the mean values for soluble solids content (SSC), titratable acid (TA), pH value and SSC-TA ratio, and the statistically significant differences were found. Principal component analysis was performed on the four chemical compositions, and obvious clustering tendencies for different varieties were found. These results showed that NIR combined with BP-ANN can discriminate strawberry of different varieties effectively, and the difference in chemical compositions of different varieties strawberry might be a chemical validation for NIR results.
Asunto(s)
Fragaria/clasificación , Redes Neurales de la Computación , Espectroscopía Infrarroja Corta , Calibración , Análisis Discriminante , Frutas , Análisis de los Mínimos Cuadrados , Modelos Teóricos , Análisis de Componente Principal , Máquina de Vectores de SoporteRESUMEN
Intense spikes caused by extraneous factors are sometimes found in the interferograms of open-path Fourier transform infrared (OP/FT-IR) measurements. Those spikes result in dominant oscillations in the corresponding spectra and make the true spectral features indistinguishable from the noise. Three techniques were designed to remove the spikes: replacing the affected region by zeroes; grafting the data from the corresponding region on the other side of the centerburst ("homografting"); and grafting the data from the same region on the same side of the centerburst from a neighboring interferogram ("heterografting"). Results showed that all three techniques were effective to remove the spikes, but at the cost of introducing more noise into the corresponding spectrum. The performance of the heterograft technique was the best in terms of noise reduction. The factors affecting the noise level of spectra computed from spike-removed interferograms were also explored. A procedure was proposed to minimize the effect of spikes on OP/FT-IR spectra, and a Matlab program with a graphical user interface is available upon request to facilitate the implementation.
RESUMEN
The use of multiple calibration sets in partial least squares (PLS) regression was proposed to improve the quantitative determination of NH(3) over wide concentration ranges from open-path Fourier transform infrared (OP/FT-IR) spectra. The spectra were measured near animal farms, where the path-integrated concentration of NH(3) can fluctuate from nearly zero to as high as approximately 1000 ppm-m. PLS regression with a single calibration set did not cover such a large concentration range effectively, and the quantitative accuracy was degraded due to the nonlinear relationship between concentration and absorbance for spectra measured at low resolution (1 cm(-1) and poorer.) In PLS regression with multiple calibration sets, each calibration set covers a part of the entire concentration range, which significantly decreases the serious nonlinearity problem in PLS regression occurring when only a single calibration set is used. The relative error was reduced from approximately 6% to below 2%, and the best results were obtained with four calibration sets, each covering one quarter of the entire concentration range. It was also found that it was possible to build the multiple calibration sets easily and efficiently without extra measurements.
RESUMEN
The automated quantification of three greenhouse gases, ammonia, methane, and nitrous oxide, in the vicinity of a large dairy farm by open-path Fourier transform infrared (OP/FT-IR) spectrometry at intervals of 5 min is demonstrated. Spectral pretreatment, including the automated detection and correction of the effect of interrupting the infrared beam, is by a moving object, and the automated correction for the nonlinear detector response is applied to the measured interferograms. Two ways of obtaining quantitative data from OP/FT-IR data are described. The first, which is installed in a recently acquired commercial OP/FT-IR spectrometer, is based on classical least-squares (CLS) regression, and the second is based on partial least-squares (PLS) regression. It is shown that CLS regression only gives accurate results if the absorption features of the analytes are located in very short spectral intervals where lines due to atmospheric water vapor are absent or very weak; of the three analytes examined, only ammonia fell into this category. On the other hand, PLS regression works allowed what appeared to be accurate results to be obtained for all three analytes.
RESUMEN
Open-path FT-IR spectra were measured while fireworks were emitting smoke and incandescent particles into the infrared beam. These conditions were designed to simulate the appearance of smoke and explosions in a battlefield. Diethyl ether was used to simulate the vapor-phase spectra of G agents such as sarin. The measured interferograms were corrected by a high-pass filter and were rejected when interfering features were of such high frequency that they could not be removed by application of this filter. The concentration of diethyl ether was calculated correctly by partial least squares regression in the absence of fireworks but significant errors were encountered when the spectra of the oxide particles were not included in the calibration set. Target factor analysis allowed the presence of the analyte to be detected even when the incandescent particles were present in the beam.
Asunto(s)
Atmósfera/análisis , Sustancias para la Guerra Química/análisis , Éter/análisis , Espectroscopía Infrarroja por Transformada de FourierRESUMEN
Open-path FT-IR spectra of low-concentration releases of diethyl ether were measured both when a glycol fog was passed into the infrared beam and when large water droplets from a lawn sprinkler were sprayed into the beam. It was shown that the glycol fog, for which the droplet size was much less than the wavelength of the infrared radiation, gave rise to a significant interference such that partial least squares (PLS) regression would only yield reasonable values for the ether concentration if background spectra in which the glycol fog was present were included in the calibration set. On the other hand, target factor analysis (TFA) allowed the presence of the ether to be recognized without precalibration. When large water droplets were present in the beam, any infrared radiation entering the droplet was completely absorbed, so that both PLS and TFA would yield accurate results.
Asunto(s)
Atmósfera/análisis , Sustancias para la Guerra Química/análisis , Lluvia , Tiempo (Meteorología) , Éter/análisis , Espectroscopía Infrarroja por Transformada de FourierRESUMEN
A theoretical investigation into the mechanism of information extraction by target factor analysis (TFA) is presented from experimental data in the form of a matrix, and the results were validated using composite spectra obtained by open-path Fourier transform-infrared (FT-IR) spectrometry. The composite spectra were generated by adding the spectral information of a target molecule with known path-integrated concentrations to the raw open-path FT-IR spectra obtained in a pristine atmosphere. Target molecules are deemed to be detected when the weighted correlation coefficient between the calculated spectrum of the analyte and its reference spectrum exceeds 0.90. The effective detection by TFA is shown to depend on the variation of their concentrations over the period of the measurement and not necessarily on the magnitude of concentration. When TFA fails to detect an analyte at high, but relatively constant, concentration that varies so little as to have low variance, blank spectra, i.e., spectra in which the analyte is known to be absent, are included in the data matrix. This procedure effectively increases the variance of the concentrations in the whole data set, and TFA detects the analyte.
RESUMEN
A technique for spectral searching with noisy data is described that improves the performance over contemporary approaches. Instead of simply calculating the correlation coefficient between the spectrum of an unknown and a series of reference spectra, greater weight is given to the more intense features in the reference spectra. The weight array, w, is given by |r|/{1 + d}, where the vector r represents the reference spectrum and the difference vector, d, contains the difference between the sample and reference data points, equal to |s-kr|, where k is a scaling factor that eliminates the effect of signal strength. By this approach, a large weight is only given to those points that have relatively high absorbance and are close to their counterparts in the reference spectrum. This technique was shown to give significantly improved performance when applied to noisy spectra of trace atmospheric components obtained by target factor analysis.
RESUMEN
Atmospheric analysis by open-path Fourier-transform infrared (OP/FT-IR) spectrometry has been possible for over two decades but has not been widely used because of the limitations of the software of commercial instruments. In this paper, we describe the current state-of-the-art of the hardware and software that constitutes a contemporary OP/FT-IR spectrometer. We then describe advances that have been made in our laboratory that have enabled many of the limitations of this type of instrument to be overcome. These include not having to acquire a single-beam background spectrum that compensates for absorption features in the spectra of atmospheric water vapor and carbon dioxide. Instead, an easily measured "short path-length" background spectrum is used for calculation of each absorbance spectrum that is measured over a long path-length. To accomplish this goal, the algorithm used to calculate the concentrations of trace atmospheric molecules was changed from classical least-squares regression (CLS) to partial least-squares regression (PLS). For calibration, OP/FT-IR spectra are measured in pristine air over a wide variety of path-lengths, temperatures, and humidities, ratioed against a short-path background, and converted to absorbance; the reference spectrum of each analyte is then multiplied by randomly selected coefficients and added to these background spectra. Automatic baseline correction for small molecules with resolved rotational fine structure, such as ammonia and methane, is effected using wavelet transforms. A novel method of correcting for the effect of the nonlinear response of mercury cadmium telluride detectors is also incorporated. Finally, target factor analysis may be used to detect the onset of a given pollutant when its concentration exceeds a certain threshold. In this way, the concentration of atmospheric species has been obtained from OP/FT-IR spectra measured at intervals of 1 min over a period of many hours with no operator intervention.
Asunto(s)
Atmósfera/análisis , Espectroscopía Infrarroja por Transformada de Fourier/instrumentación , Algoritmos , Monitoreo del Ambiente , Análisis de los Mínimos CuadradosRESUMEN
The effect of a nonlinear response of mercury cadmium telluride (MCT) detectors to photon flux is to cause a large offset and a slow variation in the zero-line of single-beam Fourier transform infrared (FT-IR) spectra, which dramatically reduce the accuracy to which strongly absorbing bands or lines can be measured. We describe a noniterative numerical technique by which the baseline offset can be corrected by adjusting the values of the maximum point in the interferogram (the "centerburst") and the points on either side. The technique relies on the presence of three spectral regions at which the signal is known to be zero. Two of these are found in all spectra, namely, the region below the detector cutoff and the high-wavenumber region just below the Nyquist wavenumber where the interferogram has been electronically filtered. In open path FT-IR measurements there are several regions where atmospheric water vapor and CO2 are totally opaque. We have selected the region around 3750 cm(-1). This algorithm is even shown to work well when the interferogram is clipped, i.e., the value at the centerburst exceeds the dynamic range of the analog-to-digital converter.
RESUMEN
A recently developed chemical shift anisotropy amplification solid-state nuclear magnetic resonance (NMR) experiment is applied to the measurement of the chemical shift tensors in three disaccharides: sucrose, maltose, and trehalose. The measured tensor principal values are compared with those calculated from first principles using density functional theory within the planewave-pseudopotential approach. In addition, a method of assigning poorly dispersed NMR spectra, based on comparing experimental and calculated shift anisotropies as well as isotropic shifts, is demonstrated.