摘要
为了便捷地对甘薯薯块进行种类识别,提出了一种基于图像颜色和纹理特征的薯种识别算法,能准确、鲁棒地根据薯块的横切面图像识别出薯块的薯种。首先,分别使用颜色直方图方法、灰度共生矩阵方法和Gabor滤波方法提取出5个薯块图像的颜色、纹理特征。然后,对该特征向量进行优化组合。最后,基于组合后的混合图像特征向量,应用BP人工神经网络对样本集进行训练、分类。试验表明,本研究提出的方法能准确地对甘薯块的种类进行识别,平均识别成功率达90%。
In the paper, a new method to recognize easily sweet potato species and robustly the cross -cutting images of tubers has been proposed based on colors and texture features. First, color histogram, Gray Level Co - oc- currence Matrix and Gabor filter have been utilized to extract the five color and texture features from tuber image. Second, the feature vector was combined together compatibly. Finally, BP artificial neural network was used to train and classify the tubers samples based on this mixed feature vector of image. The experimental results show that the proposed method can recognize the species of sweet potatoes accurately ; the average successful recognition rate is 90%.
出处
《中国粮油学报》
EI
CAS
CSCD
北大核心
2014年第11期118-122,128,共6页
Journal of the Chinese Cereals and Oils Association
基金
现代农业产业技术体系建设专项资金(CARS-11-B-18)
省重大专项浙江省旱粮农业新品种选育(2012C12902)
浙江省自然科学基金(LQ13F020013)
浙江省教育厅项目(Y201225450)
