摘要
一位新闻记者-机器人与机器学习的工作人员新闻编辑每日新闻-机器学习的新研究是一篇报告的主题。据新华社记者从成都发回的新闻报道,研究人员称:“高Re固溶透射电子显微镜(HRTEM)图像为研究材料的原子显微结构、位错形态、缺陷和相特征提供了有价值的信息。但目前对晶体材料HRTEM图像的分析和研究严重依赖人工技术。”这是费力的，容易受到主观错误的影响。我们的新闻编辑引用了西南石油大学的研究，"本文提出了一种机器学习和深度学习相结合的方法来自动划分晶体高分辨透射电镜图像中相同相位区域,通过滑动窗口遍历整个图像,计算每个窗口中快速傅立叶变换(FFT)的振幅SPECT RUM,将生成的数据转换成四维(4D)格式,并用主成分分析(PCA)对该4D数据进行估计."非负矩阵事实化(NMF)将数据分解为表示特征区域分布的系数矩阵和对应于FFT磁谱的特征矩阵，基于深度学习的相位识别能够识别每个特征区域的相位，通过对氧化锆和氧化物纳米颗粒HRTEM图像的实验，证明了该方法能够实现人工分析的一致性。
Abstract
By a News Reporter-Staff News Editor at Robotics & Machine Learning Daily News Daily News-New research on Machine Learning is th e subject of a report. According to news reporting originating from Chengdu, Peo ple's Republic of China, by NewsRx correspondents, research stated, "The High Re solution Transmission Electron Microscope (HRTEM) images provide valuable insigh ts into the atomic microstructure, dislocation patterns, defects, and phase char acteristics of materials. However, the current analysis and research of HRTEM im ages of crystal materials heavily rely on manual expertise, which is labor-inten sive and susceptible to subjective errors." Our news editors obtained a quote from the research from Southwest Petroleum Uni versity, "This study proposes a combined machine learning and deep learning appr oach to automatically partition the same phase regions in crystal HRTEM images. The entire image is traversed by a sliding window to compute the amplitude spect rum of the Fast Fourier Transform (FFT) in each window. The generated data is tr ansformed into a 4-dimensional (4D) format. Principal component analysis (PCA) o n this 4D data estimates the number of feature regions. Non-negative matrix fact orization (NMF) then decomposes the data into a coefficient matrix representing feature region distribution, and a feature matrix corresponding to the FFT magni tude spectra. Phase recognition based on deep learning enables identifying the p hase of each feature region, thereby achieving automatic segmentation and recogn ition of phase regions in HRTEM images of crystals. Experiments on zirconium and oxide nanoparticle HRTEM images demonstrate the proposed method achieve the con sistency of manual analysis."
NETL
NSTL