A liquid crystal tunable filter based shortwave infrared spectral imaging system: Design and integration

This paper presents the methodology to design and integrate a liquid crystal tunable filter (LCTF) based shortwave infrared (SWIR) spectral imaging system. The system consisted of an LCTF-based SWIR spectral imager, an illumination unit, a frame grabber, and a computer with the data acquisition software. The spectral imager included an InGaAs camera (320 × 256 pixels), an SWIR lens (50 mm, F/1.4), and an LCTF (20 mm aperture). Four multifaceted reflector halogen lamps (35 W, 12 VDC) were used to build the illumination unit. The system was integrated by a LabVIEW program for data acquisition. It can capture hyperspectral or multispectral images of the test object in the spectral range of 900–1700 nm. The system was validated by differentiating sugar from wheat flour, and water from 95% ethanol. The results showed that the system can distinguish these materials in both spectral and spatial domains. This SWIR spectral imaging system could be a potential useful tool for nondestructive inspection of food quality and safety.     

A high-resolution airborne four-camera imaging system for agricultural remote sensing

This paper describes the design and testing of an airborne multispectral digital imaging system for remote sensing applications. The system consists of four high resolution charge coupled device (CCD) digital cameras and a ruggedized PC equipped with a frame grabber and image acquisition software. The cameras are sensitive in the 400 to 1000 nm spectral range and provide 2048 × 2048 active pixels with 12-bit data depth. A 24 mm lens is attached to each camera via an F to C mount adapter, resulting in an imaging size of 0.63 times the flight altitude. The four cameras are equipped with blue (430–470 nm), green (530–570 nm), red (630–670 nm), and near-infrared (NIR) (810–850 nm) bandpass interference filters, respectively, but have the flexibility to change filters for desired wavelengths and bandwidths. The cameras are arranged in a quad configuration and attached to adjustable mounts that facilitate aligning the cameras horizontally, vertically, and rotationally. The image acquisition software allows the synchronized black-and-white band images from the cameras to be viewed on the computer monitor in any one of the four modes: a quad, one band image at a time, a normal color composite, or a color-infrared (CIR) composite. The band images are refreshed continuously to allow the operator to selectively save images with correct areas of interest. The selected four-band composite image is saved as a tiff file and consecutive images can be saved in 1-s intervals. A band-to-band alignment procedure based on the first- and second-order polynomial transformations was presented to further align the four band images. The system performed well in both stationary and airborne testing conditions. Airborne images obtained from agricultural fields, rangelands, and waterways demonstrate that this system has potential for monitoring crop pest conditions, mapping invasive weeds and assessing natural resources.     

A liquid crystal tunable filter based shortwave infrared spectral imaging system: Calibration and characterization

Calibration is a critical step for developing spectral imaging systems. This paper presents a systematic calibration and characterization approach for a liquid crystal tunable filter (LCTF) based shortwave infrared (SWIR) spectral imaging system. A series of tests were conducted to validate the linearity of the system output, measure the field of view of the spectral imager, increase the system spectral sensitivity, test the spatial and spectral resolution of the system, evaluate the system stability and image distortion, and reduce the spectral noise of the system output. Results showed that the system had an angle of view of 6.98° and a spatial resolution of 158 μm. The spectral sensitivity of the system was corrected by controlling the camera exposure time and gain, which increased the signal to noise ratio of the system by 16.5%. Test results also verified the system spectral accuracy and linearity (r > 0.999). The system output was proven to be stable and image distortion was not perceivable. Results of calibration tests indicated that this system satisfied the design criteria in both spatial and spectral domains. The calibration methods presented here are applicable to the LCTF-based spectral imaging systems in other applications.     

A high quality low-cost digital microscope minirhizotron system

An easily reproducible system is presented for minirhizotron image acquisition based on a digital microscope and entirely built with low-cost components. The system weighs 2.28 kg, is connected to a portable computer through USB, and allows collection and storage of high-quality digital images without other components or power sources. Settings at 25× magnification provide pixel sizes of 25 and 12 μm respectively with a medium pixel density microscope (640 × 480 pixel), and a high pixel density microscope (1280 × 1024 pixel). This kind of system coupled with recent developments of robotics and automatic image analysis for minirhizotron frames allows to envisage fast, low-labour and low-cost root investigation methods with a high degree of automation.     

Using vegetation index and modified derivative for early detection of soybean plant injury from glyphosate

Glyphosate is a non-selective, systemic herbicide highly toxic to sensitive plant species. Its use has seen a significant increase due to the increased adoption of genetically modified glyphosate-resistant crops since the mid-1990s. Glyphosate application for weed control in glyphosate-resistant crops can drift onto an off-target area, causing unwanted injury to non-glyphosate resistant plants. Thus, early detection of crop injury from off-target drift of herbicide is critical in crop production. In non-glyphosate-resistant plants, glyphosate causes a reduction in chlorophyll content and metabolic disturbances. These subtle changes may be detectable by plant reflectance, which suggests the possibility of using optical remote sensing for early detection of drift damage to plants. In order to determine the feasibility of using optical remote sensing, a greenhouse study was initiated to measure the canopy reflectance of soybean plants using a portable hyperspectral image sensor. Non-glyphosate resistant soybean (Glycine max L. Merr.) plants were treated with glyphosate using a pneumatic track sprayer in a spray chamber. The three treatment groups were control (0 kg ae/ha), low dosage (0.086 kg ae/ha), and high dosage (0.86 kg ae/ha), each with four 2-plant pots. Hyperspectral images were taken at 4, 24, 48, and 72 h after application. The extracted canopy reflectance data was analyzed with vegetation indices. The results indicated that a number of vegetation indices could identify crop injury at 24 h after application, at which time visual inspection could not distinguish between glyphosate injured and non-treated plants. To improve the results a modified method of spectral derivative analysis was proposed and applied to find that the method produced better results than the vegetation indices. Four selected first derivatives at wavelength 519, 670, 685, and 697 nm could potentially differentiate crop injury at 4 h after treatment. The overall false positive rate was lower than the vegetation indices. Furthermore, the derivatives demonstrated the ability to separate treatment groups with different dosages. The study showed that hyperspectral imaging of plant canopy reflectance could be a useful tool for early detection of soybean crop injury from glyphosate, and that the modified spectral derivative analysis had a better performance than vegetation indices.     

The development and application of an electrical impedance spectroscopy measurement system for plant tissues

For the impedance measurement system for plant tissues, a stable current source, the magnitude of which may be affected by the capacity of the operational amplifier, loading and high-frequency noise. Hence, this application note describes the use of a second-generation current conveyor (CCII) to improve the electrical circuit, plus a voltage buffer is used to significantly reduce the loading effect. A personal computer is integrated with LabVIEW software for data collection, analysis and processing. This forms a highly-effective measurement system which improves the portability of the measurement instrument. Within the frequency range 100 Hz–1 MHz, two groups of precision resistance-capacitance are adopted: 49.98 kΩ, 1.01 nF; 49.98 kΩ, 9.8 nF to execute the measurement simulation. The results of simulation measurement show that all the errors are within 5%, and the relative standard deviation within 0.30%, and these confirm that the system is highly accurate and precise in this implementation. Carrot is chosen as the target object for practical evaluation, for which the electrical impedance spectrum is a circular arc with its circular center below the real axis. Results for carrot are presented.     

Application of Response Surface Methodology to systematically optimize image quality in computer tomography: A case study using fresh chestnuts (Castanea spp.)

The objective of this study is to describe a method that will systematically and efficiently obtain a model of computerized tomographic (CT) scanning factor levels, which returns optimized high quality CT images. Chestnut (Castanea spp.) two-dimensional CT images were used to describe this optimization procedure, considered to be a critical step in the development of a fast, nondestructive technique, capable of assessing fresh internal quality attributes and components of chestnuts, and other agricultural commodities. Response Surface Methodology (RSM), using a three-factor, three-level Box–Behnken statistical design and digital image processing, were used to optimize the factors affecting image quality, which include X-ray voltage, current, and slice thickness. Response variables representing image quality were digitally and automatically inferred from fresh chestnut image Signal to Noise Ratio, Teflon® cylinder reference volume accuracy, quality assurance (QA) High Contrast Spatial Resolution phantom, and QA Low Contrast Detectability phantom. Second-order RSM prediction models for each response variable reflected a combined maximized CT image quality at a voltage, current, and slice thickness equal to 120 kV, 170 mA, and 2.5 mm respectively. The experiment yielded optimal chestnut CT images that can accurately reflect internal decay of fresh chestnuts with an overall accuracy rate equal to 96%, taking as reference data the of Subjective Quality Rating of five trained chestnut experts.     

Measurement of knot width in CT images of Norway spruce (Picea abies [L.] Karst.)—Evaluating the accuracy of an image analysis method

X-ray computed tomography (CT) is an effective scanning technology for imaging inner roundwood features. It is assumed that using this technology for log inspection prior to the first conversion could further optimise raw material utilisation. If the information obtained by CT scanning of a log shall be exploited in an industrial application, automated interpretation of CT images is a necessity. An image analysis method for knot extraction and measurement was tested on CT images of Norway spruce (Picea abies [L.] Karst.) produced by a prototype of a dedicated roundwood CT scanner. Measurement accuracy was evaluated through a comparison of the computer measurements obtained on single CT slices to reference measurements obtained manually on physical log cross-sections that corresponded to the images treated. In a linear regression established from 119 measurement pairs 68% of the variation of the manually measured knot width could be explained by the variation of the computer measurements. Bias and root mean squared error (RMSE) were 1.7 mm and 4 mm, respectively. Hence the measurements were generally in accordance, but an overestimation of knot width by the computer vision method was observed. An algorithm for removing bright sapwood areas from the CT images—necessary as a preparation for knot segmentation—has been found to fail at images of logs with partly dried sapwood; this problem was identified as the primary reason for the loss of 28% of the measurements originally to be acquired and for the occurrence of erroneous measurements. It should be investigated whether a modification of this algorithm could enhance the performance and accuracy of the method.     

Multispectral imaging system with interchangeable filter design

The design and calibration of a three-band image acquisition system was reported in this paper. The prototype system developed in this research was a three-band spectral imaging system that acquired two visible-band images and one NIR image simultaneously. This was accomplished by using a three-port imaging system that consisted of three identical monochrome cameras, an optical system, and three interchangeable optical filters. Spectral reflectance from an object was collimated by a front lens, and split in three ways by a cold mirror and beamsplitter: a cold mirror reflects 90% visible light and transmits 80% NIR light. The visible light was again split identically into two directions by an additional beamsplitter. Focusing lenses then projected each image onto its corresponding sensor. By incorporating an interchangeable filter design, the imaging system can measure any two visible spectral bands that range between 400 nm and 700 nm, and one NIR band that ranges between 700 nm and 1000 nm without any complicated manufacturing process. In order to co-register the three images, a system-specific calibration algorithm was developed that compensates for lens-sensor geometric misalignments.The prototype imaging system and the system calibration algorithm were tested and evaluated for image alignment accuracy. The imaging system acquired three-band images of 3D objects with 0.39 pixel misalignment error on average.     

Evaluation of three-dimensional accelerometers to monitor and classify behavior patterns in cattle

Cattle behavior is potentially a valuable indicator of health and well-being; however, natural movement patterns can be influenced by the presence of a human observer. A remote system could augment the ability of researchers, and eventually cattle producers, to monitor changes in cattle behavior. Constant video surveillance allows non-invasive behavior monitoring, but logging the movement patterns on individual animals over long periods of time is often cost prohibitive and labor intensive. Accelerometers record three-dimensional movement and could potentially be used to remotely monitor cattle behavior. These devices collect data based on pre-defined recording intervals, called epochs. Our objectives were to (1) determine if accelerometers can accurately document cattle behavior and (2) identify differences in classification accuracy among accelerometer epoch settings. Video-recorded observations and accelerometer data were collected from 15 crossbred beef calves and used to generate classification trees that predict behavior based on accelerometer data. Postural orientations were classified as lying or standing, while dynamic activities were classified as walking or a transition between activities. Video analysis was treated as the gold standard and logistic regression models were used to determine classification accuracy related to each activity and epoch setting. Classification of lying and standing activities by accelerometer illustrated excellent agreement with video (99.2% and 98.0% respectively); while walking classification accuracy was significantly (P < 0.01) lower (67.8%). Classification agreement was higher in the 3 s (98.1%) and 5 s (97.7%) epochs compared to the 10 s (85.4%) epoch. Overall, we found the accelerometers provided an accurate, remote measure of cattle behavior over the trial period, but that classification accuracy was affected by the specific behavior monitored and the reporting interval (epoch).     

基于Arduino和LabVIEW的水稻种植环境数据采集系统的应用

为了实现水稻种植环境的自动监测,给水稻种植和管理带来便利,设计了一种基于Arduino和LabVIEW的水稻种植环境参数的监测系统。首先介绍了监测系统的功能和结构框图,随后阐述了其硬件和软件的设计与实现。由传感器前端触杆与大气和土壤接触,采集大气的温湿度、土壤的温湿度和pH,数据采集模块将获得的信号通过串口上传至上位机。上位机LabVIEW对采集的数据进行存储、图形实时显示及处理、分析,实现实时、便捷地检测水稻环境的温度、湿度、pH变化情况。实践表明,该设计能够经济、高效地实现数据采集,可用于实时环境状况的快速监测,具有一定的参考价值和实用性。     

基于虚拟仪器技术的果园生态环境数据采集系统设计与实现

本文首先简要介绍了虚拟仪器技术及其开发平台LabVIEW,然后阐述基于LabVIEW虚拟仪器技术对果园生态环境进行数据采集的系统设计原理和实现方法。该系统采用研华ADAM-4017数据采集模块,通过串口进行数据传递,实现了果园生态环境信息的计算机自动采集、显示和存储,为果园生产管理提供及时准确的园区信息。     

Comparison of segmentation algorithms for cow contour extraction from natural barn background in side view images

Computer vision techniques are a means to extract individual animal information such as weight, activity and calving time in intensive farming. Automatic detection requires adequate image pre-processing such as segmentation to precisely distinguish the animal from its background. For some analyses such as gait analysis, a side view perspective is recommended. When using a side view angle however, the background is more difficult to control – moving objects, such as other animals may negatively impact successful image segmentation. The objective of this research was to evaluate five different background segmentation algorithms on side view images when taken against a static background (a solid transportable wall) and a dynamic background (open air, without a wall).The experiments were conducted on a commercial robotic-milking dairy farm in Israel with a herd size of 70 Israeli Holstein cows. A side view image of cow’s gait was recorded after milking when the cows exited the milking area and returned to the cowshed. From the recording database, a random selection was made of 35 frames containing a static background (solid wall) and 20 frames containing a dynamic background (natural barn environment with other cows).Five segmentation algorithms were chosen and adapted from literature to extract the cow shape from the image. The output of three algorithms gave the cow’s full body shape two identified only the contour of the cow’s body. The algorithms were compared on their ability to correctly identify the cow’s back contour line. The performance of each algorithm was quantified by comparing its outputs to a golden standard of manually labelled cow pixels in the image.The introduction of a physical wall behind the cows (static background) significantly improved the foreground segmentation results (Mean Absolute Error (MAE) = 6.7 ± 5.7 pixels vs. 19.7 ± 9.1 pixels). The fourth algorithm, based on an edge detection on the background difference frame, gave the best cow back contour line segmentation results (b₀ = −0.4 ± 15.5 and b₁ = 1.00 ± 0.07). The fifth algorithm which is based on consecutive frame differences was less accurate than the other four methods which are based on the background frame differences (MAE = 16.0 ± 5.9 pixels vs. 4.1 ± 2.2 pixels, 4.3 ± 2.2 pixels, 5.6 ± 2.8 pixels and 3.7 ± 1.4 pixels respectively for the other four algorithms). The results show that the applied algorithms were not robust enough to work on side view images with dynamic backgrounds.     

Hyperspectral detection of rice damaged by rice leaf folder (Cnaphalocrocis medinalis)

The reflectance from rice (Oryza sativa L.) leaves and canopy damaged by rice leaf folder (RLF), Cnaphalocrocis medinalis (Guenée) was studied at the booting stage in order to establish a monitoring method for RLF based on hyperspectral data. The results showed that reflectance from rice leaves significantly decreased in the green (530–570 nm) and near infrared (700–1000 nm) regions, and significantly increased in the blue (450–520 nm) and red (580–700 nm) regions as the leaf-roll rate of rice increased. Reflectance from rice canopy significantly decreased in the spectral regions from 737 to 1000 nm as the infestation scale of RLF increased, and the most correlation appeared at 938 nm. Seven spectral regions 503–521, 526–545, 550–568, 581–606, 688–699, 703–715, and 722–770 nm at leaf-level, and one region 747–754 nm at canopy-level were found to be sensitive bands to exhibit the damage severity in rice by RLF. The position of the red edge peak remarkably moved to blue region, and the amplitude and area of the red edge significantly decreased when rice leaves were severely infected by RLF. Thirty-eight spectral indices at leaf-level and 29 indices at canopy-level were found to be sensitive to leaf-roll rate and infestation scale in rice, respectively. The linear regression models were built to detect the leaf-roll rate (0.0–1.0) and infestation scale (0–5) in rice using leaf- and canopy-level reflectance data. The root mean square error of the model was only 0.059 and 0.22 for the leaf-roll rate and infestation scale, respectively. These results suggested that the hyperspectral reflectance was potential to detect RLF damage severity in rice.     

A longitudinal study of the effect of time on the matching performance of a retinal recognition system for lambs

While there is strong evidence supporting retinal vascular pattern as a distinctive marker for sheep, it would be advantageous to get an insight into its robustness; in other words, to determine whether retinal recognition of young animals (lambs) can reach as good a matching performance as the one demonstrated for adult sheep. To this aim, a longitudinal study was devised to observe the evolution of matching scores (ms) of lamb retinal images (n = 38) acquired from 1 to 22 weeks after birth. It was observed that four lamb retinas (out of 38) underwent slight curving of one or two secondary arteries, which ceased by the time they were 6–8 weeks old. However, this slight artery curving did not affect matching performance. A random effects statistical model demonstrated that lamb age had an effect (P < 0.01) on the matching scores produced using this commercially available retinal recognition system. As lambs grew older (larger eyes) and they became easier to handle, retinal images of progressively better quality could be obtained in a more consistent way; and thus matching scores increased from an average of 86 at the first week of life, up to an average of 96 by week 8. Dunnett simultaneous tests of means indicated that no further improvement in matching score took place once lambs were at least 6–8 weeks old, meaning that the retinal image quality became by then optimal and consistent. Although the variable retinal image quality of younger lambs (1–4 weeks old) caused a reduction in matching score, they did not lead to false non-matches in any case (considering a cut-off matching score of 70 for acceptance of a positive match). Therefore, the results of these trials have shown that, with the available technology, retinal images can be used as a robust biometric marker of lambs from 1 week of age.     

Acceleration of CT reconstruction for wheat tiller inspection based on adaptive minimum enclosing rectangle

Tiller number is highly correlated with grain yield in wheat. Traditional observation of wheat tiller number is still manual. Previously, our group developed a high-throughput system for measuring automatically rice tillers (H-SMART) based on X-ray computed tomography (CT), providing high accuracy for measuring rice tillers. However, the time-consuming reconstruction, which is necessary to generate tomographic images, limits the throughput improvement of system as well as the CT potential for the real-time applications. In order to accelerate the reconstruction process, we present an adaptive minimum enclosing rectangle (AMER) method to reduce the number of reconstructed pixels from the full field of view (FOV) and apply parallel processing using Graphics Processing Unit (GPU). The reconstruction time and speedup with different methods were discussed. Compared to the AMER method, GPU technique improved reconstruction with a higher speedup of approximately 200 times. And the speedup with AMER method was determined by two factors: area ratio of AMER and FOV, and the longest distance between the vertices of the AMER and the rotation center. Besides reconstruction, tiller identification could also be accelerated by AMER. Moreover, the tiller measurement accuracy did not decrease. With the combination of AMER and GPU, the entire tiller inspection time for a pot-grown plant was reduced from about 11870 ms to less than 200 ms. In sum, the optimized method met the requirement of real-time imaging and expanded CT application in plant phenomics and agriculture photonics.     

Field monitoring support system for the occurrence of Leptocorisa chinensis Dallas (Hemiptera: Alydidae) using synthetic attractants,Field Servers,and image analysis

To realize effective insect counting in pheromone traps set in remote sites, a remote monitoring and image processing system based on a sensor network system of “Field Servers” has been developed, and two practical methods based on image analysis using this system has been proposed. This system has been employed to monitor the occurrence of the rice bug, Leptocorisa chinensis, in rice paddy fields as a means of reducing the burden of manual insect counting work.A Field Server with a high-resolution digital camera was installed near the pheromone trap for close monitoring. The image data and other monitoring data such as temperature were sent via wireless LAN and the Internet every 5 minutes. A remote management system for the Field Server, located about 7.5 km from the experimental field, managed data collection and analyzed the data to provide useful information on insect count. An image analysis algorithm based on a background differencing technique has been developed to support counting L. chinensis by implementing an image-processing module in the remote management system. The image-processing module provides three analysis functions: cropping, subtracting, and binarizing the target image.One method is to filter extraneous image data containing no observed target insects (end-members) on the pheromone trap. In this method, the difference between collected image data and the reference image data was calculated, and the total number of pixels whose value was greater than a threshold value for the difference result (number of white pixels) was used for filtering. This method managed to maintain Sensitivity at 100% during the experiment. Accuracy was observed to be 89.1% on average. Using this method, the time spent looking at extraneous image data without L. chinensis can be reduced by 85%.The other method for reducing labor in counting involves estimating the number of end-members automatically using a partial image area that is cropped to focus on a low-noise area, permitting easy analysis. With this method, the image data was analyzed using the first method, and the entire number of end-members was estimated using the number of white pixels and a pixel value equivalent to one end-member. The results of this method correspond reasonably closely to the results obtained by manual counting. The correlation coefficient for the daily occurrence rate was 0.974 and that for the hourly rate was 0.916.     

基于LabVIEW与SQL Server的心电数据库设计

介绍基于LabVIEW的心电采集系统的后台数据库设计，利用LabSQL工具包建立LabVIEW与SQL Server数据库的连接，并利用SQL Server设计专门的心电数据库，将数据库技术与LabVIEW平台结合，解决大规模心电数据带来的管理问题。文中介绍的系统可以实现数据库与LabVIEW之间的连接和数据传输，并将采集到的心电数据存储在心电数据库中，完善心电采集系统的功能，为心电数据的管理和后续利用提供了便利。     

Line-scan hyperspectral imaging system for real-time inspection of poultry carcasses with fecal material and ingesta

In poultry processing plants, fecal material and ingesta are the primary source of carcass contamination with microbial pathogens. The current practice of the poultry inspection in the United States is primarily human visual observations. Since the visual inspection is becoming more challenging in poultry processing plants adopting high-speed lines, a rapid sorting system could significantly improve the detection and monitoring of carcasses with surface fecal material and ingesta. As a result, we developed a prototype line-scan hyperspectral imaging system configured as a real-time multispectral imaging subsystem for online detection of surface fecal material and ingesta. Specifically, we integrated a commercially available off-the-shelf hyperspectral image camera into the system with two line lights and a custom software program for real-time multispectral imaging. The bottleneck of the imaging system was the data acquisition. For that reason, a multithreaded software architecture was designed and implemented not only to meet the application requirements such as speed and detection accuracy, but also to be customizable to different imaging applications such as systemic disease detection in the future. The image acquisition and processing speed tests confirmed the system could operate to scan poultry carcasses in commercial poultry processing plants. The fecal detection algorithm was based on the previous research using different hyperspectral imaging systems. A new carcass detection and image formation algorithm was developed to allow existing image processing and detection algorithms reusable without any modifications. Sixteen chicken carcasses and four different types of fecal and ingesta samples were used in a study to test the imaging system at two different speeds (140 birds per minute and 180 birds per minute) in a pilot-scale poultry processing facility. The study found that the system could grab and process three waveband images of carcasses moving up to 180 birds per minute (a line-scan rate 286 Hz) and detect fecal material and ingesta on their surfaces. The detection accuracy of the system varied between 89% and 98% with minimum false positive errors (less than 1%), depending on tested detection algorithms. Therefore, these findings provide the basis of not only a commercially viable imaging platform for fecal detection but also a single poultry inspection system for multiple tasks such as systemic disease detection and quality sorting.     

Analysis of hyperspectral scattering images using locally linear embedding algorithm for apple mealiness classification

Hyperspectral scattering images between 600 nm and 1000 nm were acquired for 580 ‘Delicious’ apples for mealiness classification. A locally linear embedding (LLE) algorithm was developed to extract features directly from the hyperspectral scattering image data. Partial least squares discriminant analysis (PLSDA) and support vector machine (SVM) were applied to develop classification models based on the LLE, mean-LLE and mean spectra algorithms. The model based on the LLE algorithm achieved an overall classification accuracy of 80.4%, compared with 76.2% by the mean-LLE algorithm and 73.0% by the mean spectra method for two-class classification (i.e., mealy and nonmealy) coupled with PLSDA. For the SVM models, the LLE algorithm had an overall classification accuracy of 82.5%, compared with 79.4% by the mean-LLE algorithm and 78.3% by the mean spectra method. Hence, the LLE algorithm provided an effective means to extract hyperspectral scattering features for mealiness classification.