基于BP神经网络的葡萄干分级技术的研究

Research of Classification of Raisin Based on BP Networks

为了适应农产品机器视觉分级技术的需要,通过分析葡萄干的特征参数组合对BP网络识别率的影响,确定最能反映葡萄干形态特征的参数作为等级识别的依据。为此,以BP算法为基础,采用改进的BP算法,对网络进行训练;同时,建立了基于BP神经网络葡萄干分级鉴定模型。试验结果表明,该网络分级效果较好,其平均分级准确率达到93%。

In order to adapt to the primary products request of the machine vision grading technology. This article through analysis raisin characteristic parameter combination＇s influence of BP network recognition rate, and definite the parameters which could reflect the shape features perfectly. Taking the BP algorithm as the foundation, and the network was trained by improved BP; a model for accreditation raisin was presented on the basis of BP networks. Experiments verifies that this BP networks be of better results. Its classification average accuracy rate could reach 93%.