基于BP神经网络的机器人图像压缩技术  

Image Compression Technology of Robot Based on BP Neural Network

作　　者：蔡钟山 CAI Zhongshan(Fujian Special Equipment Inspection and Research Institute,Fuzhou 350008,China)

机构地区：[1]福建省特种设备检验研究院,福州350008

出　　处：《机电工程技术》2020年第5期101-102,132,共3页Mechanical & Electrical Engineering Technology

摘　　要：视觉机器人通过网络通信实时将海量的图像视频采集并上传,图像数据容量大,容易造成图像传输不及时或者出错。针对该问题,提出一种基于BP神经网络的视觉机器人图像压缩技术,采用图像客观评价体系中的图像相似度(SSIM)和峰值信噪比(PSNR),用于评价图像质量。经MATLAB仿真表明,基于LM训练算法的BP神经网络能够对图像有效压缩,选用合适的隐含层神经元数目,图像压缩后的图像与原图像的相似度可达到92%,峰值信噪比达到36.3 dB,压缩后图像质量能够保证,具有一定的工程应用价值。The visual robot needs to collect and upload massive images and videos in real time through network communication, and the large image data capacity can easily cause the image transmission to be not timely or wrong. To solve this problem, a visual robot image compression technique based on BP neural network was proposed, which used image similarity(SSIM) and peak signal-to-noise ratio(PSNR) in the image objective evaluation system to evaluate image quality. The MATLAB simulation shows that the BP neural network based on LM training algorithm can compress the image effectively, select the appropriate number of hidden layer neurons, and the similarity between the image and the original image can be achieved after image compression 92%, the peak signal-to-noise ratio(SNR) reaches 36.3 dB, and the image quality is guaranteed after compression.

关 键 词：LM算法 BP神经网络 图像压缩 SSIM PSNR 

分 类 号：TP391[自动化与计算机技术—计算机应用技术]