基于高光谱成像提取苹果糖度与硬度最佳波长

发布时间：2018-06-03 04:25

本文选题：高光谱成像 + 苹果　；参考：《发光学报》2017年06期

【摘要】：利用高光谱成像技术提取可同时检测苹果糖度与硬度的最佳波长。首先双面采集苹果的高光谱图像,获取亮度相近感兴趣区域(ROIs)的反射波形,采用二阶导数结合标准正态变量(SD+SNV)的方法平滑波形,测试ROIs的糖度与硬度;之后采用连续投影算法(SPA)提取两项指标的特征波长,根据特征波长的分布提出二次连续投影算法,结合波形集特征与两次投影结果确定不同取样面的最佳波长;最后采用遗传算法开发神经网络(GA-BP)建立预测模型,双面取样波长(543 nm和674 nm)效果最优,糖度相关系数(R)为0.847 6,均方误差(MSE)为3.32;硬度R为0.793 8,MSE为9.6。结果表明,相同波长信息可以检测苹果糖度与硬度。
[Abstract]:The optimum wavelength for simultaneous detection of sugar and hardness of apple can be extracted by hyperspectral imaging technique. Firstly, the hyperspectral images of apple were collected on both sides, and the reflection waveforms of the region of interest with similar brightness were obtained. The second derivative method combined with the standard normal variable (SD SNV) was used to smooth the waveforms, and the saccharification and hardness of ROIs were tested. Then the continuous projection algorithm (spa) is used to extract the characteristic wavelengths of the two indexes. According to the distribution of the characteristic wavelengths, the quadratic continuous projection algorithm is proposed, and the optimum wavelength of different sampling surfaces is determined by combining the waveform set features with the two projection results. Finally, the genetic algorithm was used to develop the neural network (GA-BP) to establish the prediction model. The two-sided sampling wavelength was 543 nm and 674nm), the correlation coefficient of sugar degree was 0.847 6, the mean square error (MSE) was 3.32, and the hardness R was 0.793 8 nm and 9.6nm. The results showed that the same wavelength information could be used to detect the sugar content and hardness of apple.
【作者单位】：沈阳农业大学信息与电气工程学院;辽宁广播电视台;辽宁广播电视传输发射中心;
【基金】：辽宁省大型仪器设备共享服务项目(LNDY201501003) 沈阳市大型仪器设备共享服务专项项目(F15-166-4-00)资助~~
【分类号】：O657.3;TS255.7

【相似文献】