摘要
金属断口图像疲劳条带间距是影响疲劳应力计算的重要因素。电子显微镜拍摄的金属断口显微图像较为清晰,主要噪声影响是断口表面的灰尘对提取疲劳条带造成的干扰。图像分块后,根据子块的灰度均值、方差和方向图区分条带块和噪声块,并利用噪声块邻近的条带块沿条带纹理方向滑动对噪声块进行修复,实现去噪。对去噪后的图像Gabor滤波、二值化、细线化、轮廓规整等操作,最后对每条连续曲线进行直线拟合。实验结果表明,通过上述预处理方法可以得到较为理想的疲劳条带图像轮廓。
Fatigue striation width in metal fracture image is an vital factor which affects the results of retroestimating fatigue stress. Metal fracture images captured by electron microscope are clear. Main noise originates from the dust on fracture surface, which is the disturbance when fatigue striation is extracted. After metal fracture image is blocked, each image block is classified into striation block and noisy block according to gray mean, variance and directional field. Neighboring striation blocks glided along the direction of striation texture so that noise blocks are restored to realise de-noising. The image after de-noised is Gabor filtered, binarized, thinned, then contour is optimized. Finally each continuous curvy is line fitted. Experiments show that the above pre-processing method could extract fatigue striation contour effectively.
出处
《计算机工程与设计》
CSCD
北大核心
2009年第3期640-642,共3页
Computer Engineering and Design
基金
国家自然科学基金项目(60475002)
航空科学基金项目(2008ZC56001)
