一种快速剔除伪分枝的作物行骨架提取算法

作物行骨架线的提取是机器视觉导航的基础,准确提取作物行骨架也是精准施药系统中一个至关重要的研究方向。为了克服传统骨架提取算法中的背景单一、存在较多的冗余分支及不连续等缺点,以农田作物行为研究对象,提出一种形态学细化和伪分支剔除相结合的实用型骨架提取算法。首先对采集到的作物行原图像通过灰度、滤波、阈值分割操作使其转化为二值图像;然后将二值化后的作物行图像先细化为单像素宽度的骨架线,再采用端点追踪法追踪伪分支骨架,而后剔除追踪的毛刺或无关枝杈,保证了骨架的单一性和圆滑性,提高了作物行检测的精度。通过与拓扑细化法和最大圆盘骨架提取算法比较,本文算法不但在去除冗余骨架的同时能保持自身良好的拓扑性和稳定性,而且能去除多余的毛刺状分支,同时表明该算法具有较强的稳定性和抗干扰能力。

A Crop Row Skeleton Extraction Algorithm Based on Morphological Thinning and Pseudo Branch Elimination

The skeleton extraction of the crop rows is the basis of machine vision navigation and it is a vital research in precision spraying system.In order to overcome the shortcomings of the traditional framework extraction algorithm, such as a single background, more redundant branches and not continuous, etc.This paper to crop study proposed an algo-rithm, which combined morphological thinning and pseudo branch elimination to extract crop rows skeleton.Firstly, the crop rows that were collected through gray, filtering, threshold segmentation operation to convert it to binary contour map.Then the crop row image of binarization was refined to a single pixel width line.And then the endpoint tracking method is used to track the pseudo branch skeleton.At the same time, the tracing burr or unrelated branches were dele-ted, which ensured the singleness and smooth of the skeleton and improved the accuracy of crop rows detection.Com-pared with the traditional method based on topology refinement and maximum disk skeleton extraction algorithm, the algo-rithm that is proposed in this paper the algorithm could not only remove the redundant skeleton, but also could maintain its good topology, and it also shows that the algorithm has good stability and anti-jamming capability.