基于改进的灰色关联度的根系图像边缘检测

目的实现对植物根系形态基本参数的计算和分析。方法针对阵列分布内窥式图像获取技术采集的根系图像,提出一种改进的灰色关联度的边缘检测算法。该算法基于灰色关联分析理论,采用变权的邓式关联度模型,利用Sobel算子的2个模板作为参考序列,选取像素的八邻域分量值形成比较序列,通过两类序列之间的关联度实现边缘检测。结果仿真结果表明该算法与传统的边缘检测算子相比,能够较准确地检验出有用的根系边缘信息。结论基于Sobel算子的变权关联度的根系边缘检测算法有效地提高了边缘检测效果,具有一定的抗噪性能。

Root Image Edge Detection Based on Improved Gray Correlation Degree

In order to calculate and analyze the basic parameters of plant root system, an improved gray correlation degree edge detection algorithm was proposed by acquisition technique for array distributed endoscopic image. Based on the grey relational analysis theory, the algorithm adopted the variable weight model, took the two templates of Sobel operator as the reference sequence, selected the eight neighborhood component values of pixels to form a comparison sequence and realized the edge detection by the correlation degree between the two kinds of sequences. The simulation results showed that the proposed algorithm was able to accurately test the useful information of the root edge compared with the traditional algorithm. In conclusion, the edge detection algorithm of variable weight correlation degree based on Sobel operator can effectively improve the edge detection effect, and has a certain anti noise performance.