%0 Journal Article %T 应用高光谱图像拟合铜绿假单胞菌的生长 %A 顾欣哲 %A 孙晔 %A 王文雪 %A 王振杰 %A 屠康 %A 潘磊庆 %J 南京农业大学学报 %D 2016 %R 10.7685/jnau.201509010 %X [目的] 本文旨在开发一种新颖、快速、无损的方法模拟铜绿假单胞菌的生长。[方法] 利用高光谱图像技术获取铜绿假单胞菌在营养琼脂平板上培养0、12、24、36和48 h的图像和光谱参数,采用部分短波近红外波长范围的平均光谱值、高光谱图像的掩模图形和480~1000 nm波段内光谱值的第1主成分共3种方法来模拟铜绿假单胞菌的生长。[结果] 对初始浓度为102和104 CFU？mL-1的假单胞菌所建立的模型训练集决定系数(R2)为0.9701~0.9932,平方误差和(SSE)为0.00003~0.08061,验证集R2为0.8057~0.9542,SSE为0.00028~0.10461。3种基于高光谱图像特征参数建立的模型与基于菌落计数法建立的模型之间的相关系数均超过0.900,说明基于高光谱图像参数建立的假单胞菌生长模型符合其实际生长情况。基于部分波长范围的平均光谱值和基于高光谱图像的掩模图形建立的拟合模型比第1主成分得分建立的模型与数据拟合效果更好,具有更高的准确性。同时,对480~1000 nm波段范围内的光谱值进行主成分分析(PCA),发现2种浓度的铜绿假单胞菌的5个检测阶段只有部分样品点重叠,相互之间可以区分。[结论] 高光谱图像技术可以用来对铜绿假单胞菌的生长状况进行区分和模拟,为进一步对冷鲜肉中的铜绿假单胞菌检测奠定基础。</br>[Objectives] The purpose of this paper is to develop a new, rapid and nondestructive method to simulate the growth of Pseudomonas aeruginosa.[Methods] Generally, the study of P. aeruginosa in cold fresh is to make qualitative and quantitative description by mathematical model and set up its growth dynamic model. However, these models are based on calculating the number of living bacterium, testing the metabolites and destructive testing. It’s time-consuming and energy-wasting time. Meanwhile, it needs a lot of time to get results. This article used a hyperspectral imaging system(HIS)to acquire the image and spectral response of P. aeruginosa inoculated on nutrient agar plate at 0, 12, 24, 36 and 48 h, and then to extract characteristic parameters to simulate the growth of P. aeruginosa. Three methods, including the mean of the spectral response of short-wave infrared spectroscopy, mask pattern of hyperspectral imaging and the score of PC1 of the whole spectral range of 480-1000 nm, were used to simulate the growth situation of P. aeruginosa.[Results] The results indicated that the coefficients of determination(R2)of simulation models based on three methods for testing datasets were high with 0.9701 to 0.9932, and the sum square error(SSE)was in a range of 0.00003-0.08061. For validation datasets, R2 of simulation models based on three methods was from 0.8057 to 0.9542 and SSE was in a range of 0.00028-0.10461. The correlation coefficients(r) between the HIS parameters and colony forming units of P. aeruginosa were all beyond 0.900, which indicated the growth models of P. aeruginosa based on hyperspectral image parameter were in line with the actual growth situations. The models based on the mean of the spectral response of short-wave infrared spectroscopy and mask pattern of hyperspectral imaging had a better fitting effect and higher accuracy than the scores of PC1 of the whole spectral range of 480-1000 nm. In addition, the growth situation of P. aeruginosa can be %K 高光谱图像 %K 铜绿假单胞菌 %K 生长模拟 %K 区分< %K /br> %K hyperspectral imaging %K Pseudomonas aeruginosa %K simulation models %K discrimination %U http://nauxb.njau.edu.cn/oa/darticle.aspx?type=view&id=201603023