基于机器视觉的变量机械系统的关键技术

简要介绍了应用于变量机械的GPS技术与机器视觉技术两种方式的优缺点;重点阐述了光源照明、动态图像采集(CCD摄像机、图像采集卡)、图像处理和快速准确的执行机构等基于机器视觉的变量机械的关键技术的发展现状;最后扼要介绍了机器视觉在田间导航、农田喷洒等变量机械中的应用及展望。     

连接大米籽粒图像的自动分割算法研究

分析了单个大米籽粒和连接大米籽粒的轮廓特征,提出了一种基于曲率的分割算法。该算法能够有效地反映边界点尖锐程度的曲率及其方向,根据曲率大小和曲率方向能准确判别图像中大米籽粒是否接触并快速找到分割点。用最短距离配对法对3种接触的大米籽粒进行分割。算法对判别籽粒是否接触的准确率达99%以上,对接触籽粒的准确分割率达95%以上。     

水泵吸水池流场PIV分析

水泵吸水池内部的流场很复杂，特别是在吸入管的周围，分布着十分复杂的涡。随着流量的变化，这些涡的位置和强度也相应地变化，这也是吸水池影响水泵系统性能的直接原因。一旦吸水池内部结构确定，影响其流场的因素就是来流情况。本文使用PIV技术对不同来流情况下吸入管附近的流场进行测量，从而得到该吸水池对不同来流的适应性。     

基于移动溯源与图像分析的茶叶品级鉴定方法研究

为解决茶叶在质量追溯时防伪和品质鉴定问题, 以六安瓜片为研究对象, 提出一种结合移动计算和图像分析的茶叶质量追溯方法。该方法的实现以 Android操作系统为平台, 通过扫描茶叶质量追溯二维码查询溯源信息。在此基础上, 利用智能终端拍照获取完整的单片茶叶正面照片并上传至后台服务器。后台编写的 MATLAB图像处理程序对采集的图像进行预处理和特征参数提取, 将提取的颜色和形状共 7项特征参数作为参数集合, 运用 SVM分类器对茶叶进行分类, 最终将结果反馈到溯源客户端。测试结果证明, 该方法能快速帮助消费者对茶叶品级进行初步鉴定。     

基于弹性动量深度学习神经网络的果体病理图像识别

为了实时预警果蔬病害和辅助诊断果蔬疾病,实现无人值守的病虫害智能监控,设计了深度学习神经网络的果蔬果体病理图像识别方法,基于对网络误差的传播分析,提出弹性动量的参数学习方法,以苹果为例进行果体病理图像的识别试验。结果表明,该方法召回率为98.4%;同其他同源更新机制相比,弹性动量方案能显著改善学习网络的果蔬病害识别准确率;其收敛曲线平滑,5 h时耗能实现收敛,对不同数据集也有良好泛化性能。     

农业机械视觉导航基准线识别研究进展

导航基准线识别是农业机械自主导航的关键技术。为此,在参考相关文献的基础上,对目前国内的相关研究进行较为系统的分析和总结;介绍了近年来的国外研究进展,最后根据现有研究状况展望了今后的研究方向。     

Image edge detection algorithm in fuzzy domain

An image edge detection algorithm in fuzzy domain is proposed, which combines adaptive fuzzy enhancement and multi direction fuzzy morphology to detect the edges of fuzzy image. The adaptive fuzzy enhancement method enhances the fuzzy image within blocks with sliding windows to avoid losses of the real edges resulting from enhancing with single threshold for the whole image and lead strong adaptive ability to image region variance. The multi direction fuzzy mathematical morphology operates on the enhanced fuzzy image with structure elements of multiple different directions to extract the real edges with directionality and restrain non directional noise. Experiments show the algorithm can detect fuzzy image edge effectively with strong antinoise ability.     

Fast circle measurement of industrial CT images based on existence probability

The general methods in reverse engineering cannot measure the inside of close parts. Traditional methods like Capture Geometrical Inside (CGI) would destroy the part. We used the industrial computer tomography (ICT) method to obtain accurate measurements of close parts. Based on the acquired computer tomography images, the size of an internal part was obtained through image measurement and subsequently edited, amended, and dimensioned in CAD software. First, the position and the pixel size of the circle were determined based on existence probability. Next, the actual size of one pixel was obtained based on the image size and the purview diameter. Finally, the actual circle size was calculated by multiplying the pixel size and the actual size of one pixel. A real part was measured using the method. The results show that the relative error of the measured circle radius and the actual size is less than 0.5% and the measuring time is several seconds.     

Image denoising based on adaptive graph regularization

Adaptive regularization can select different parameters based on the features of local areas in an image, which can differentiate the edges and noise in an image flexibly. An adaptive graph regularization is proposed based on graph spectral theory and adaptive regularization, which uses the Non local means to generate the weighting function of graph. The adaptive graph regularization equation is used to filter the noisy image. Simulation results show that the proposed method can effectively remove the noise and is superior to other graph theory based partial differential equation methods.     

Automatic quality assessment of fingerprint image

Based on multiple influencing factors, a new method of automatic fingerprint image quality evaluation is proposed for improving the success rate of automation fingerprint identification system (AFIS). At first, the original image’s local texture, global texture, available size and dry or wet condition are regarded as quality impact factors, local texture quality score is calculated by local gradient correlation, and then the last three factors’ quality scores are obtained by block computation thought. Then, with different influence weights, the above four impact factors are linked together to assess image quality synthetically. Finally, effect of partial impact factor is adjusted to correct the final quality score. FVC2004DB2_B is used for algorithm testing. The results suggest that this method can reasonably classify fingerprint image into 5 grades and the precision can achieve 97.5%, and that shows the method is helpful to success rate of AFIS.