基于深度神经网络的鸟类图像识别系统设计

Design of Bird Image Recognition System Based on DNN

目前机器视觉和生物图像检测绝大多数都是针对人脸识别,鲜有对于野生动物物种识别的研究。生物特征识别技术主要包含图像预处理,特征提取,特征选择和分类器设计。系统通过卷积神经网络CNN和OpenCV结合数字图像处理技术,对于野生鸟类进行种类识别,通过大量正负样本训练,提取出不同鸟类的特征文件,利用深度学习网络对于不同图像进行识别,表明该鸟类图像识别效率较高,结构精巧,具有一定的推广使用价值。

At present,most of machine vision and biological image detection are directed at face recognition,and there is little research on wildlife species identification.Biometric recognition technology mainly includes image preprocessing,feature extraction,feature selection and classifier design.The system uses the convolutional neural network CNN and OpenCV combined with digital image processing technology to identify the wild birds,and through a large number of positive and negative samples training,extract the feature files of different birds,and use the deep learning network to identify different images,indicating the bird image recognition has high efficiency,exquisite structure,and certain promotion and use value.