基于光切法的葡萄硬枝嫁接切削面粗糙度图像检测算法

发布时间：2018-06-10 06:01

本文选题：嫁接 + 算法　；参考：《农业工程学报》2017年04期

【摘要】：为了实现葡萄硬枝嫁接苗木切削面粗糙度的检测,该文基于光切法测量原理,搭建了切削面粗糙度图像检测系统,研究了特征提取的图像检测算法。为获取较长的取样长度,采用了图像拼接技术,并提出了一种自动制取匹配模板的方法,拼接算法测试结果表明:每多拼接一幅粗糙度特征图像,运行时间平均增加1.104 s,取样长度平均增加1 131.77μm;采用了模糊集合理论对拼接后的粗糙度特征图像进行灰度变换,可以有效保证图像分割后单侧边缘的完整;采用了人机交互的方式对粗糙度特征二值图像像素进行区域操作,可以滤除因切削面自身含有的导管腔、管胞腔而导致的缺陷轮廓,从而提高粗糙度计算的准确度;提出了一种逐列遍历图像提取单侧边缘的方法,通过对单侧边缘进行计算,可以得到粗糙度高度参数Ra与Rz的值。将该粗糙度图像检测算法与基恩士VK-X200形状测量激光显微系统进行了粗糙度检测对比试验,结果表明,该文提出的粗糙度图像检测算法测得Ra的相对误差为6.73%,在测量误差允许范围内,该文基于光切法测量原理的图像检测算法,用于检测葡萄硬枝嫁接苗木切削面粗糙度,具有较高的精度和良好的可行性,为进一步研究切削参数对切削面粗糙度以及对苗木嫁接成活率的影响提供了技术支撑。
[Abstract]:In order to detect the roughness of grafted grape plantlets, based on the measuring principle of light cutting method, an image detection system for cutting surface roughness was set up, and an image detection algorithm for feature extraction was studied. In order to obtain longer sampling length, image stitching technique is adopted, and a method of automatically making matching template is proposed. The test results of stitching algorithm show that: one roughness feature image is stitched with each other. The average running time and sampling length are increased by 1.104 s and 1 131.77 渭 m respectively, and the gray level transformation of the feature image of roughness after stitching is carried out by using fuzzy set theory, which can effectively guarantee the integrity of the edge of one side after image segmentation. The method of human-computer interaction is used to operate the pixels of the binary image of roughness feature, which can filter out the defect profile caused by the tube cell and the tube cavity itself, thus improving the accuracy of roughness calculation. In this paper, a method of extracting unilateral edges from traversal images is presented. The values of roughness height parameters Ra and Rz can be obtained by calculating the edge of one side. The roughness detection algorithm is compared with the VK-X200 shape measurement laser microscope system. The results show that, The relative error of Ra measured by the roughness image detection algorithm proposed in this paper is 6.73. Within the limits of the measurement error, the image detection algorithm based on the principle of light cutting method is used to detect the roughness of grafted grape plantlets. It has high precision and good feasibility, which provides technical support for further study on the effect of cutting parameters on cutting surface roughness and survival rate of seedling grafting.
【作者单位】：中国农业大学工学院;
【基金】：北京市自然科学基金资助项目(6152012)
【分类号】：S663.1;TP391.41

【相似文献】