基于图像增强与边缘检测技术的蔗芽识别

Sugarcane Bud Recognition Based On Image Enhancement and Edge Detection Technology

准确识别蔗芽是蔗芽位姿调整,进行蔗种定向种植的前提。以大津法,形态学运算来提取单个蔗种图像;通过基于模糊集和浮雕效果的图像增强算法和基于小波变换模极大值的边缘检测算法,凸显蔗芽区域;基于图像乘法和形态学运算等方法去除蔗芽轮廓与蔗芽区域噪声,识别并提取出蔗芽区域。以“桂糖44号”甘蔗取种为例,试验选取160个预切蔗种段,对320幅蔗种(含蔗芽面200幅,不含面0幅)图像进行识别,结果表明:上述方法对含蔗芽面和不含蔗芽面检测的准确率分别为92.5%和89.2%,蔗种蔗芽识别的平均准确率为91.3%。以上研究基于图像处理技术识别蔗芽,为调整蔗芽位姿进行蔗种定向种植提供技术支持。

Obtaining the accurate position of the sugarcane buds plays an important role in the adjustment of the position of the sugarcane buds in the directional planting of sugarcane seeds. In order to accurately detect the exact position of sugarcane buds, a sugarcane buds recognition method based on image processing and edge detection is proposed. The maximum between-class variance method and mathematical morphology processing were used to extract the image of a single sugarcane bud. Through the image enhancement method based on fuzzy set and relief algorithm and edge detection algorithm based on wavelet transform modulus maxima, the area of sugarcane buds was highlighted. Based on image multiplication and mathematical morphology processing methods, the contour of the sugarcane buds and the noise of the sugarcane buds area were removed, and the sugarcane buds area was extracted and identified. Taking the sugarcane seed collection of “Guitang 44” as an example, 160 sugarcane seed segments after artificial cutting were selected to identify the results of 320 sugarcane seed images(including 200 sugarcane buds and 120 without sugarcane buds). It shows that the accuracy of the method for detecting the surface with and without cane buds is 92.5% and 89.2%,respectively, and the average accuracy of sugarcane buds is 91.3%. This research is based on image processing and detection technology to identify sugarcane buds, and provides technical support for adjusting the position and posture of sugarcane buds for directional planting of sugarcane seeds.