CAJ | 학술논문

为实现鸟类物种识别和自动化观测，针对鸟的图像的自动提取和识别过程中的问题，开展了一系列的研究工作。首先对鸟的彩色图像进行二值化，接着采用 Log 算法对鸟的轮廓进行提取，然后对鸟进行初步颜色特征提取，对特征颜色较明显的鸟进行粗分类，再接着对特征色不太明显的鸟类进行基于灰度共生矩阵算法的纹理特征提取，最后对提取的纹理特征进行 BP 神经网络分类，最终达到对鸟的识别。实验结果表明，平均识别正确率达到70%以上。
위실현조류물충식별화자동화관측，침대조적도상적자동제취화식별과정중적문제，개전료일계렬적연구공작。수선대조적채색도상진행이치화，접착채용 Log 산법대조적륜곽진행제취，연후대조진행초보안색특정제취，대특정안색교명현적조진행조분류，재접착대특정색불태명현적조류진행기우회도공생구진산법적문리특정제취，최후대제취적문리특정진행 BP 신경망락분류，최종체도대조적식별。실험결과표명，평균식별정학솔체도70%이상。
In order to realize the identification and automatic observation of the bird species, a series of research work based on the automatic extraction and identification has being developed. To achieve the recognition of the birds, employing binarization of the bird’s color image first, and then, using the Log algorithms to extract the outline of the birds. Next, according to the preliminary extraction of the color features of the birds, processing coarse classification to the birds and extracting the texture features based on gray level occurrence matrix to the birds which are not obvious on the color characteristics. Finally, classifying the texture features on the BP neural network. The experimental results show that the average recognition accuracy above 70%.