Multi object tracking is one of the hotspots of computer vision research. Tracking-by-detection is a common approach to multi-object tracking. With the development of machine learning, especially of deep leaning method, the basis for a tracker becomes much more reliable. The method proposed in this paper is based on detection with convolutional neural network and tracking with hierarchical clustering. Experimental evaluation shows that the proposed method achieves overall competitive performance at high frame rates.

群养饲喂模式下猪群有聚集在一起的习性,特别是躺卧时,当使用机器视觉跟踪监测猪只时,图像中存在猪体粘连,导致分割困难,成为实现群猪视觉追踪和监测的瓶颈.根据实例分割原理,把猪群中的猪只看作一个实例,在深度卷积神经网络基础上建立PigNet网络,对群猪图像尤其是对粘连猪体进行实例分割,实现独立猪体的分辨和定位.PigNet网络采用44层卷积层作为主干网络,经区域候选网络(Region proposal networks,RPN)提取感兴趣区域(ROI),并和主干网络前向传播的特征图共享给感兴趣区域对齐层(Region of interest align,ROIAlign),分支通过双线性插值计算目标空间,三分支并行输出ROI目标的类别、回归框和掩模.Mask分支采用平均二值交叉熵损失函数计算独立猪体的目标掩模损失.连续28 d采集6头9.6kg左右大白仔猪图像,抽取前7d内各不同时段、不同行为模式群养猪图像2 500幅作为训练集和验证集,训练集和验证集的比例为4:1.结果 表明,PigNet网络模型在训练集上总分割准确率达86.15％,在验证集上准确率达85.40％.本文算法对不同形态、粘连严重的群猪图像能够准确分割出独立的猪个体目标.将本文算法与Mask R-CNN模型及其改进模型进行对比,准确率比MaskR-CNN模型高11.40个百分点,单幅图像处理时间为2.12s,比Mask R-CNN模型短30 ms.

In land-use classification of hyperspectral remote sensing (RS) images, traditional classification methods often experience large amount of datasets and low efficiency. To solve these problems, a fast machine-learning method, the extreme learning machine (ELM) algorithm, was introduced. However, basic use of the ELM usually encounters problems of unstable classification results and low classification accuracy. Hence, in this paper, optimization methods for ELM-based RS image classification were mainly discussed and applied to solve the bottleneck problems. From the three perspectives of ensemble learning, making full use of image texture features, and deep learning, three classification optimization methods have been designed and implemented. The results show that: 1) To some extent, all the three methods can achieve a balance between classification accuracy and efficiency, i.e., they can maintain the advantage of ELM algorithm in classification efficiency and speed while have better classification accuracy; 2) The image texture feature optimization method (LBP-KELM) solves the problem of unsatisfactory classification results experienced by the ensemble learning optimization method (Ensemble-ELM) and further improves classification accuracy. However, the classification results are sensitive to the type of dataset; and 3) Fortunately, the optimization method combined with deep learning (CNN-ELM) can meet the application needs of multiple datasets. Furthermore, it can also further improve classification accuracy.

High temperature coal tar pitch is a kind of high quality raw material to produce carbon materials. However, coal tar pitches without refining treatment was not suitable to produce such high quality carbon materials. Actually, refining treatment is an important way to adjust the molecular weight distribution, aromatic structure, and aliphatic side chain structure of coal tar pitch. Refining treatment is a precondition to produce carbon materials which is easy to graphitize with coal tar pitch as the raw material. In this study, four kinds of refined pitches numbered as RP-1, RP-2, RP-3 and RP-4 have been obtained by two kinds of preparation methods with the medium pitch and modified pitch as the raw materials, respectively. The refined pitches have been studied by Fourier Transform infrared spectroscopy (FTIR) and curve-fitting analysis in order to gain additional information on the comparation of these four refined pitches. The curve-fitted data provide quantitative evidence of aromaticity(Iar) length of aliphatic chain(CH3/CH2), distribution of OH groups, and oxygen-containing functional groups with different refined pitches. The results have showed that: these four kinds of refined pitches has a larger aromatic condensation degree. RP-3 has the highest aromaticity index of 0.9. RP-4 has the Index of the branched chain of 0.07, which means that RP-4 have a long aliphatic chain. The distribution of OH groups in refined pitches was significantly different. The results can provide a theoretical support for the selected raw materials in the preparation of carbon/graphite materials.

In this study, we propose a method for correcting the adverse effects produced by the cur vature of fruit objects in images acquired by cameras in machine vision systems. The areas near the edge are darker in acquired images than in the centre, which results in many difficulties for subsequent analyses. In this paper, the fruit object was considered as a Lambertian surface. The light intensity was analysed and the height and normal on fruit surface was deduced based on the shape-from-shading (SFS) algorithm. The geometric correction factors were calculated and the adverse effects of light intensity were corrected on the fruit surface. The proposed method was evaluated on a test set of four types of fruit. The results show that the non-uniformity of the greyscale value on the fruit surface fell by 35.5% after correction, and the ref lectance in the central area of fruit is similar to that of the peripheral areas when using the proposed method. The experiments prove that our method allowed homogenisation of the greyscale level of the pixels belonging to the same class, regardless of where they are on the fruit surface, which will facilitate subsequent classification tasks.

Detecting immature fruit in groves provides a promising benefit for growers to plan application of nutrients and estimate their yield and profit prior to harvesting. The goal of this study was to develop a robust algorithm to detect and count immature citrus fruit in images of the tree canopy. Images were all taken in low natural light conditions with a flashlight, and the green component of the colour images was used for further analysis. Local intensity maxima were detected and local binary pattern (LBP) features around them were extracted as an input of an ensemble classifier-RUSBoost. The positive predictions were considered as candidates and the hierarchical contour maps around them were extracted and fitted with Circular Hough Transform. The fitted circles were predicted as fruit targets if its radius were in a predetermined range. The algorithm was evaluated with a test set of 25 images, achieved 80.4% true positive rate and 82.3% precision rate, and F-measure was 81.3%. The good performance of occlusion tolerance of the proposed method was mainly coming from the robust LBP texture descriptor and hierarchical contour analysis (HCA) which used the pattern of light intensity distribution on fruit surface. This study proposed an innovative method to detect green fruit in images of trees only by using texture and intensity distribution. (C) 2018 IAgrE. Published by Elsevier Ltd. All rights reserved.

针对传统大学物理实验预习中存在的学生对实验原理理解不深,无法接触实验仪器等问题,提出了基于虚拟仿真实验预习和全开放预习实验室的虚实结合混合式大学物理实验预习模式,该模式依托虚拟仿真实验预习和全开放预习实验室进行大学物理实验预习,将线上虚拟仿真实验预习和线下全开放预习实验室学习有机融合,充分发挥了远程网络学习和本地学习的优势,提高了学生自主预习兴趣、预习质量和预习效果.

为提高玉米果穗发育程度检测的自动化程度与精度,提出一种基于机器视觉技术的测量方法.在随机森林机器学习算法的基础上构造秃尖、干瘪和籽粒区域的识别模型.该模型由多个独立同分布的弱分类器构成,对输入的训练样本进行列和行两个方向上的随机采样.比较随机森林模型和决策树模型的分类效果可知随机森林模型有效避免了过拟合和局部收敛现象的产生,并具有良好的推广能力.为确定最优的弱分类器数目,选择弱分类器个数为训练样本数量的1/80、1/40、1/20、1/10、1/5、1/4时分别构建随机森林分类器.研究结果表明,当随机森林中弱分类器个数为训练样本数量的1/20时,模型的识别率与稳定性最好.然后,以最优的随机森林模型作为分类器构建玉米果穗不同发育程度自动检测方法.试验结果表明,各区域长度测量的准确性均在95％以上,测量速度可达30个/min以上.

Accurate crop-load estimation is very important for efficient management of nutrients and harvest operations. Current machine vision techniques for crop-load estimation have achieved only limited success mostly due to partial occlusion, shape irregularity, varying illumination and multiple sizes. Detecting immature green fruit is a more challenging task for similar color of fruit and background. The key starting point of this paper for detecting immature citrus fruit was the observation that the light distribution on citrus fruit follows a general pattern in which the light intensity decreases with the distance from a local maximum due to specular reflection. Immature citrus fruit detection was achieved by detecting this pattern with concentric circles or parts of circles. This pattern was proposed with the maximally stable extremal region (MSER) method and validated by hierarchical contour analysis (HCA) which was the first proposed in this paper. The images were captured by a color camera under low natural light conditions with a flashlight, and the green component of the color images was used for further analysis. After smoothing the whole image by Weiner filter, the regions of interest (ROIs) in the image were extracted by the method of MSER. The ROIs detected by MSER were those whose support was nearly the same over a range of thresholds, so the regions on citrus fruit were detected by MSER for the pattern that the light intensity decreases stably and gently with the distance from a local maximum. However, many regions on leaves and background were also detected as ROIs and should be excluded in the next step. A novel algorithmic technique was proposed to remove these regions on background, and this method was named as the HCA. Firstly, shape analysis was used for each ROI and only those ROIs were considered as valid if the shape was nearly circular. Secondly, multiple levels of contours around each valid ROI were extracted and fitted with the circular Hough transform (CHT). Lastly, multiple fitted circles would be merged into one if their most parts were overlapped together, this step was called circle merging and the merged circles were considered as the last detected citrus fruits. The algorithm was tested on a testing dataset with 20 images and achieved the recall rate of 81.2% and the precision rate of 83.5%. The processing time of the proposed method was 3.70 s totally on each image, on average, in which 0.57 s was used for MSER detection and 3.13 s was used for HCA. The result showed that the proposed method can detect green citrus fruit in a very difficult and challenging scene with so many fruits in one image and extensive partial occlusion. The good performance of partial occlusion tolerance of the proposed method in this paper is mainly due to that the proposed HCA doesn't use the shape of outer contour of fruit, but uses multiple concentric contours which come from the pattern of light intensity distribution on fruit surface. The research framework in this paper can give a novel thought on other green fruit detection besides citrus fruit.

This paper presented an automatic segmentation method to detect defects on the surface of citrus fruits eroded by diseases or pests based on circularity gradient threshold. A visible imaging system was built and citrus fruits were shot by this system. The chromatic aberration map of green and blue components was obtained, and a segmentation method with a circularity gradient threshold was used to detect defects from healthy regions on fruit surface. The erosion coefficient was used to quantify erosion degree by diseases or pests. The segmentation results showed that by using circularity gradient threshold segmentation method, the missing alarm rate is increased by 0.19%, but the false alarm rate is decreased by 4.44% versus global Otsu threshold method on average. The algorithm can detect defects on fruit surface accurately and integrally, and it has important potential in fruit grading, control of diseases and pests.

Fruit quality inspection techniques play a very important role in the production and consumption of fruits.In the field of quality non-destructive inspection and grading for fruits,the light-based techniques using optical properties of fruit products were widely used as one of the most practical and the most successful techniques.Quantitative understanding of light interaction with fruits is critical to designing better optical systems for inspection of food quality.In this paper,a fruit model consisted of two layer tissues was developed using Monte Carlo simulations to explore the light transport process and properties in the pome fruits,such as apples and mandarins,which were used as the thin-skinned and thick-skinned fruits respectively.The simulation results obtained are based on the assumption that the light source is a Gaussian beam at the wavelength 808 nm.This paper reports that the effects of skin thickness on light transmission characteristics in fruit tissues,including diffuse reflectance,transmittance,absorptivity,penetration depth etc.The inspection efficiency of flesh tissues was also demonstrated.The results indicated that the transmittance and the penetration depth decreases with the fruit skin increasing.As for the absorbed energy density,the fruit skin tissues have the wider distribution at the radial distance than the fruit flesh tissues.The absorbed energy density always tended to decrease with the inside depth of the fruit tissues increasing,especially decreased more apparently at the radial direction.The diffuse reflectance at the radial distance from 0.2to 1.2cm decreased with the decreasing of fruit skin,however it showed the inverse relationship in the radial distance range from 1.2to 4.0cm,the diffuse reflectance decreases with the increasing fruit skin.This paper proposed that the interaction between light and fruits skin in transmission or reflective approach,should be considered for developing optical techniques of non-destructive fruit quality inspection.And it provided a theoretical basis for designing more efficient optical detection device,including how to confirm the light source power,size and position of the detector,etc.It has very important significance for fruit quality inspection by optical techniques.

A method based on color information and contour fragments was developed to identify citrus fruits in variable illumination conditions within tree canopy, in order to guide the robots for harvesting citrus fruits. The color properties of fruit targets within citrus-grove scene were analyzed, a preliminary segmentation method was put forward by fusing the chromatic aberration information and normalized RGB model. The set of contour fragments was constructed by detecting the significant edges of chromatic aberration map and the corners within these edges. The valid subset was chosen out by three indicators of every fragment: length, bending degree, and concavity or convexity. The combination analysis was done for these valid contour fragments, and the ellipse fitting was used for every subset of valid fragments to recover the occluded fruits. The partial order relationship was derived based on the distribution of the edge within the overlapped area. The results showed that occluded fruits were effectively recovered under natural outdoor light conditions using the proposed method, and the relative error was 5.27%. The partial order relation of fruit targets provide key cues for path planning of harvesting robot.

This paper had proposed an automatic segmentation method to detect defects on citrus surface eroded by diseases and pests based on circularity threshold segmentation. A visible imaging system was built and citrus fruits were shot by this system. The chromatic aberration map of GB components was obtained, and a circularity threshold was used to separate defects from healthy regions on fruit surface. The erosion coefficient was used to quantify the erosion degree by diseases or pests. With respect to global Otsu segmentation method, the performance improvement of circularity threshold method in this paper was up to 15.58%. The algorithm can detect defects on citrus surface accurately, rapidly, and it had important significance in citrus grading, control of diseases and pests.

遮挡是自然场景中普遍存在的问题,为在变化光照条件下正确检测出自然环境中的树上成熟水果目标,从而为全天候的机械采摘提供运动参数,研究了基于彩色信息和目标轮廓整合的树上遮挡柑橘检测方法。在对自然光照条件下的可见光彩色图像进行颜色特征分析的基础上,建立了利用R-B色差图融合归一化RGB颜色空间的方法,对树上水果目标区域进行了初分割。然后提取R-B色差图的主边缘构造边缘片段集,根据边缘片段长度、弯曲程度以及凹凸性来选择有效边缘片段集,对每个有效边缘片段进行拟合,最后根据水果形状知识选择出有效拟合目标椭圆。根据对不同光照条件和遮挡程度的场景处理的结果表明,所提算法能有效恢复出树上存在遮挡的水果目标,最后遮挡轮廓恢复结果的相对误差为5.34%。

目前检测技术在自动化生产过程中越来越重要,为了实现工业现场中投影仪壳体标签的自动检测功能,提出了一种基于图像传感器的数字图像处理技术检测方法.采用工业摄像头替代目视作为传感器;利用Otsu阈值分割处理图像,将标签区域分为字符标签区域和非字符标签区域两类,两部分缺陷分开判断,互不干扰.字符标签区域利用模板匹配技术进行缺陷检测;非字符标签区域提取面积、重心、方向三个特征量,与标准模板数据进行特征匹配.实验测试表明:检测系统采集到的产品标签图像清晰,缺陷检测算法正确率达到95％以上,可实现工业自动化生产过程中投影仪壳体标签缺陷的自动化检测.

Detection infestations caused by insects in grain are important control measures for ensuring storage longevity, seed quality and food safety. The efficiency of the continuous wave terahertz imaging method to detect infestations caused by insects in wheat kernels was determined in this study. A continuous wave terahertz experimental setup was designed for recording of THz images corresponding to different infestations caused by different life stages of insects. The experimental results indicate that the absorbance is generally highest for un- infested wheat kernels and decreased at later growth stages from THz pseudo-color images. Our study intended to demonstrate how the method of continuous wave Terahertz imaging could be applied to detect Infestations Caused by Insects in Grain.

Artificial vision systems are powerful tools for the automatic guiding of fruit harvesting robots, a novel method based on chromatic aberration map and luminance map is developed to identify citrus fruits with highlight within a tree canopy. Twenty images of citrus-grove scene under direct sunlight are taken, the color properties of target objects are analyzed. First, parts of citrus fruits from background are segmented from background by thresholding the CAM (chromatic aberration map), then the highlight region of citrus fruits could be detected correctly from the tree canopy by thresholding the LM (luminance map), at last the citrus fruits can be detected integrally by fusing the segmented results of CAM and LM. The results showed that the fruits under direct sunlight can be segmented wholly using this algorithm, the detection accuracy by fusion method is up to 86.81%, and the false alarm rate of fusion method is 2.25%.

In order to guide the robots for harvesting citrus fruits, method based on normalized RGB model and chromatic aberration map was developed to detect citrus fruits with shadow within tree canopy. Several images of natural citrus-grove scene were photoed, and the color properties of target objects were analyzed. A rule for segmenting citrus fruits from background was put forward by fusing the segmented results of the normalized R channel map and the chromatic aberration map of R and B channels. The results show that the fruits with shadow can be detected integrally using the proposed method.

In the context of construction of research universities,it's very important that the college physics experiment of college of agriculture and forestry implement centralization and open teaching.It can give full play to the fundamental role of college physics experiment as a basic common course.Based on the characteristics of teaching in college of agriculture and forestry,a centralization and open teaching management system for college physics experiment was designed.The application of the system confirmed the experiment teaching system that it worked and assessment methods achieved diversity.The resources of physics experiment teaching centre have the maximum efficiency owing to the application of the system.It plays an important role in training the comprehensive quality,the spirit of innovation and the ability of practice of students.

This paper focuses on the detection of citrus fruits in the tree canopy under variable illumination and different degree occlusion. We applied a novel segmentation method to detect the visible parts of fruits by fusing the segmentation results of chromatic aberration map, normalized RGB model, and illumination map. This fusion method can detect the highlights, shadows and diffuse zones of fruit targets. The 3-D surface topography of the visible parts of fruits were recovered by the classical algorithm of shade from shading, the fruit targets were recovered by sphere fitting using these point cloud data, and the valid ones were chosen out by validity check. The results showed that the occlusion zones of targets were effectively recovered under various light conditions integrally using the proposed method.