2020-07-22
材料的腐蚀和老化往往伴随着结构、成分和形貌的改变。通过图像采集设备（数码相机、扫描电镜、声扫描显微镜以及红外照相机等）可以获得相应的数字图像。图像识别方法实际是通过各种图像采集设备对材料的失效、老化、缺陷或局部腐蚀进行表征，提取用于定性或半定量评价的特征参数。
研究室夏大海老师近十年来一直从事钝性金属腐蚀损伤机制及防护基础方面的研究工作，已经在国内外学术刊物上发表SCI论文100余篇，H指数18。现担任中国腐蚀与防护学会“青年工作委员会”及“腐蚀电化学及测试方法委员会”委员。此次，受Journal of Materials Science & Technology期刊邀请，综述了过去20多年团队在图像识别用于材料腐蚀和老化方面的研究工作，列出了常见的图像识别算法，并分别探讨了各自的适用范围及局限性。
Fig. 1. Images based on radiation from the electromagnetic (EM) spectrum and acoustic waves: (a) the EM spectrum arranged according to energy per photon, (b) an example of infrared photography of steel corrosion under an organic coating, (c) the image in (b) obtained from an optical microscope, (d) an example of X-ray tomography of corrosion pits, (e) an image of blisters on a lacquered food can, (f) a scanning acoustic microscope (SAM) image of the image in (e) clearly showing the blisters, (g) a corrosion pit observed by a scanning electron microscope.
Fig. 2. Field corrosion image of aluminum alloy exposed to an urban atmosphere: (a) image acquisition using a KH3000 digital video-microscope system (Hirox. Co. Ltd.), (b) typical corrosion morphology images of 7B04-T6 aluminum alloy samples exposed to an atmospheric environment for various times, (c) the selected region for discrete wavelet transform (DWT), (d) a one-scale DWT, (e) a two-scale DWT, (f) a three-scale DWT, (g) energy distribution in aluminum alloy samples after exposure for various times, where d = days and H, V, and D correspond to horizontal, vertical, and diagonal detailed sub-images, respectively, and E = energy value .
Fig. 3. Metallic materials corrosion image collection in the field using a portable CCD camera: (a) image collection, (b, c) corrosion images of 304 SS exposed to a marine atmosphere.
