基于图像处理技术的麦穗产量测量方法

摘要:

选用9918小麦品种，利用Matlab图像处理技术，对小麦穗头图像的纹理特征与产量的关系作了初步研究。研究结果表明：穗头图片各纹理特征参数值(灰度均值、方差、平滑度、三阶矩、一致性、熵)与穗头产量均呈显著相关；用多元线性回归方法建立的穗头图像纹理-产量数学模型，在置信度为95%时，复相关系数为0.980 7；对于产量大于、等于1.06 g的本品种穗头，用建立的模型测穗头产量，精度达15%以上的样本占

Abstract:

84.42%The relationship between the yield of a wheat spike and its image texture features was investigated with image processing technology. The spike images were obtained by a digital camera and processed with Matlab. The spike textures described with the mean value, standard deviation, smoothness, third moment, consistency and entropy were extracted based on gray level statistical properties of the spike image and the relationship between wheat spike yield and its image texture features was established by means of multiple linear-regression method. For the given breed named wheat 9918, the experimental results showed that the kernel yield of the wheat spike and its image texture features are significantly correlated with the confidence of 95% and correlative coefficient of 0.980 7. Using established model, wheat spike yield could be predicted with relative error less than 15% for 84.42% samples. 