基于人工智能的计算机图像自动识别  

Computer image automatic recognition based on artificial intelligence

作　　者：胡博 钱鑫[2] HU Bo;QIAN Xin(Information Construction and Management Center,Nanjing Normal University of Special Education,Nanjing 210000,China;Nanjing University of Aeronautics and Astronautics,Nanjing 210007,China)

机构地区：[1]南京特殊教育师范学院信息化建设与管理中心,江苏南京210000 [2]南京航空航天大学,江苏南京210007

出　　处：《电子设计工程》2024年第22期191-195,共5页Electronic Design Engineering

摘　　要：针对在复杂图像环境下,无法精准快速识别目标图像的复杂特征等问题,提出基于人工智能的计算机图像自动识别方法。在灰度级模糊C均值算法基础上,引入类贡献抑制因子与类紧密度,设计改进的快速模糊C均值算法,用于计算机图像自动分割与目标提取;同时,改进AlexNet卷积神经网络,通过卷积层、Batch Normalization层、池化层及全连接层提取并转换处理目标图像特征,得到目标图像特征向量;并通过Sortmax层处理目标图像特征向量,得到计算机图像自动识别结果。实验证明该方法能够有效自动分割并精准提取特征目标图像特征,完成计算机图像自动识别,验证了该方法在复杂图像识别中的有效性。Addressing issues such as the inability to accurately and quickly identify complex features of target images in complex image environments,the method of computer image automatic recognition based on artificial intelligence is studied.In the fuzzy C-means algorithm based on gray level,the class contribution suppression factor and class tightness are introduced,and design an improved fast fuzzy C-means algorithm for computer image automatic segmentation and object extraction.Improve the AlexNet convolutional neural network by extracting and transforming target image features through convolutional layers,batch normalization layers,pooling layers,and fully connected layers to obtain target image feature vectors.And process the target image feature vector through the Sortmax layer to obtain the computer image automatic recognition result.Experimental results have shown that this method can effectively automatically segment and accurately extract the features of the target image,completing computer image automatic recognition and verifying the effectiveness of this method in complex image recognition.

关 键 词：人工智能 计算机图像 自动识别 模糊C均值 贡献抑制因子 卷积神经网络 

分 类 号：TN929.5[电子电信—通信与信息系统]