欢迎您！东篱公司
退出
-
申报数据库
-
立项数据库
-
成果数据库
-
项目获奖数据库
中文摘要：
将核磁共振成像技术与人工神经网络理论相结合，对香梨内部褐变进行了检测。在磁共振T2加权图像中选取果核区域作为感兴趣区域，提取出反映褐变特性的10个微观纹理特征参数，建立了BP神经网络模型进行识别研究。针对BP神经网络模型存在的不足，利用遗传算法对网络模型的权值和阈值进行优化。通过验证性试验发现：对于4组香梨样本，优化后BP神经网络模型的平均正确识别率为92．50％，比未优化模型的平均正确识别率80．83％，提高了11．67个百分点；同一组香梨样本相比较，优化后模型的识别效果也均优于未优化模型，每组香梨的识别率都得到了不同程度的提高。结果表明：遗传算法优化后的BP神经网络模型具有很好的预测精度和泛化能力，可以实现香梨内部褐变的无损检测。
英文摘要：
Magnetic resonance imaging （MRI） technology and artificial neural network theory were used to discriminate the browning disease inside the fruit. Areas corresponding to the core of fragrant pear in T2-weighted image were selected to the region of interest （ROI）. Quantitative analysis of the ROI was achieved by extracting ten texture features that reflected the browning characteristics. Back propagation （BP） neural network was carried out on the statistical features to predict the internal browning of fragrant pear. Genetic algorithm （GA） was adopted to optimize the initial weights and threshold in BP neural network. For four groups of samples, the optimization model showed 92.50% accuracy in detecting the presence of browning in fragrant pear, compared with the correct recognition rate 80.83% of the non- optimization, an 11.67 percent increased. For the same group samples, the recognition results of optimized model were also better than the non-optimized model and the correct recognition rate of each group was improved to varying degrees. The result of our experiment shows that the optimized model has good predictive accuracy and generalization ability to identify the internal browning of fragrant pear.
同期刊论文项目
同项目期刊论文
Detecting Macronutrients Content and Distribution in Oilseed Rape Leaves Based on Hyperspectral Imag
Application of visible/near-infrared spectroscopy to the determination of catalase and peroxidase co
Detecting macronutrientscontent and distribution in oilseed rape leaves based on hyperspectral imagi
Determination of pigments concentration of oilseed rape (Brassica napus L.) leaves using hyperspectr
Variety recognition of chinese cabbage seeds by hyperspectral imaging combined with machine learning
Rice Seed Cultivar Identification Using Near-Infrared Hyperspectral Imaging and Multivariate Data An
Fast Detection of Peroxidase (Pod) Activity in Tomato Leaves Which Infected with Botrytis Cinerea Us
期刊信息
位置：