基于深度学习与图像处理的哈密瓜表面缺陷检测

针对传统人工检测哈密瓜表面缺陷效率低等问题,提出利用卷积神经网络(Convolutional Neural Networks,CNN)对哈密瓜表面缺陷进行快速检测.对原始图像进行主成分分析(Principal Components Analysis,PCA)、奇异值分解(Singular Value Decomposi...     

AN EFFICIENT LABELLING ALGORITHM FOR WOODPANELS SURFACE DEFECTS

This paper describes an efficient approach for labeling images using a combination of pipeline (Datacube) and (general purpose computer) processing. The output of the algorithm is coordinate list of labeled object pixels that facilitates further high level operations. It is an efficient labeling algorithm for a automatic classification of surface defects on wood boards.     

木材缺陷图象的伪彩色处理

一、引言木材缺陷的检测是评定材质高低的一个重要依据。目前,我国木材缺陷的检测基本上处于用肉眼观察确定表面的缺陷;用敲击法或凭经验来判断内部缺陷的状况。用X光透视技术检验木材内部缺陷、评定木材等级还处于实验阶段。最近,计算机数字图象处理技术的应用已发展到生物学、医学、工业等方面,并已经收到了可喜的成果。但尚未见到应用于木材检测方面的报道。本文初步介绍了用伪彩色技术处理木材缺陷图象的一些实验结果。     

Inherited erythrocyte pyruvate kinase deficiency in a beagle dog

A 1-year-old, female Beagle dog with minimal exercise intolerance was found to have a persistent, severe, and highly regenerative anemia, splenomegaly, and progressive osteosclerosis. Despite near-normal in vitro erythrocyte pyruvate kinase (PK) activity, the authors diagnosed PK deficiency by demonstrating a glycolytic block at the PK step, the lack of normal R-type PK isoenzyme, and the presence of M(2)-type PK in the animal's erythrocytes. The dam had half-normal erythrocyte PK activity, which supports an autosomal recessive mode of inheritance. We conclude from our studies that close similarities exist between erythrocyte PK deficiency in Beagle, Basenji, and West Highland White Terrier dogs and that this form of PK deficiency may be more widespread than previously thought.     

X射线扫描技术对原木内部缺陷的识别方法

对X射线的吸收能力,原木木质部分与内部缺陷部分存在不同,因此在X射线底片上形成的影像也不一样,根据一定的处理识别分析,可以对原木内部缺陷的参数进行测定。利用X射线的扫描图像,提出一种原木缺陷的识别方法。     

基于形态学重构的实木地板缺陷分割方法研究

针对传统区域生长方法中,由于噪声种子存在及种子点单步邻域搜索所导致的分割时间长、检测精度低的问题,提出基于形态学重构的实木地板在线缺陷分割方法。方法首先定义不同阈值下的两幅模版图像,其中低阈值图像用于种子优化,高阈值模版用作种子膨胀生长;通过定义腐蚀终止准则,完成低阈值图像下的缺陷骨架提取;运用"去毛刺"操作,最终实现缺陷骨架内的种子点优选;然后,运用测地膨胀,结合高阈值模版,完成板材缺陷区域的快速生长;最后,应用"孔洞填充"、"去毛刺"优化边缘,实现缺陷目标的提取。实验分别在像素512*512、256*256和128*128下进行,通过与传统区域生长方法的比较,表明方法实现了缺陷区域的准确分割,分割速度与精度能够满足地板在线分选要求。     

Study on Drying Characteristic of Chinese Fir and Poplar Plantation Wood

The drying characteristic was studied for plantation wood of Chinese fir and poplar, which are typical plantation wood of southern and northern part of China, respectively. Through 100-degree-method the drying characteristic and basic drying condition were gotten, then drying schedule was developed for practical drying, the results showed that the drying schedule is suitable for Chinese fir and poplar plantation lumber, but shrinkage is large. The recommendation was made that enough dead weight is needed to decrease shrinkage in drying process. The drying quality of the two species of lumber is good in conventional drying method.     

超声法检测钢管混凝土质量的探讨

结合湖南省益阳市资江三桥和益阳市康复南路跨线桥钢管拱内混凝土质量的超声法检测实例,利用透射的原理来判断混凝土缺陷位置及大小的方法,具有一定的实用价值,可供参考。     

基于高光谱成像技术检测脐橙溃疡

为了研究从带有不同缺陷的柑橘类水果中快速识别出溃疡果的有效方法,基于高光谱成像技术,该文提出特征波段主成分分析法及波段比算法。以脐橙为研究对象,选取包括溃疡在内常见的10类脐橙果皮缺陷果及正常果。首先,提取并分析11类果皮感兴趣区域（ROI）光谱曲线并结合主成分分析法确定5个最佳波段（630、685、720、810和 875 nm）;然后基于特征波段做主成分分析,选取第5主成分（PC-5）作为分类识别图像,识别率达到80%。为了进一步提高溃疡识别率,该文又提出采用特征波段主成分分析法与波段比算法相结合的方法,基于此算法溃疡正确识别率提高到95.4%。试验结果表明,基于高光谱成像技术可以有效地对带有溃疡病斑的脐橙分类识别。