数字细节增强技术在脉冲热成像无损检测中的应用
|
摘要
针对脉冲热成像法红外检测图像对比度低、缺陷目标边缘模糊、受不均匀照明影响大的问题,提出一种基于目标轮廓的结合数字细节增强技术与多阈值最大熵的缺陷大小定量估算方法。首先,脉冲红外检测图像经自动对比度增强算法优化的数字细节增强方法处理后,缺陷与背景间对比度显著提升,从而不均匀照明对缺陷识别效果的影响明显减弱;其次,再通过遗传算法优化的最大熵多阈值分割方法提取缺陷目标,对其进行八邻域法轮廓跟踪,以提取各缺陷区域的轮廓像素点并排序;最后,对具有一定方向的缺陷轮廓分别采用欧氏距离法和格林公式对缺陷的周长和面积进行定量估算。实验结果表明:该方法对缺陷大小进行定量估算的可行性,且数字细节增强技术可在一定程度上提高脉冲热成像检测系统的缺陷探测水平。
Pulsed thermographic image has the disadvantages of low-contrast, fuzzy-edge, and non-uniformity of illumination for the defect detection, thus, a defect determination method which combines digital detail enhancement(DDE) technology with maximum entropy multi-threshold segmentation method was proposed for the improvement of pulsed thermographic image. Firstly, the contrast between defects and the background was improved significantly after the image was processed with digital detail enhancement algorithm optimized with adaptive contrast enhancement(ACE) algorithm, and reducing the influence of illuminative non-uniformity on defect recognition. Secondly, the target defects with maximum entropy multi-threshold segmentation method optimized with genetic algorithm, and the contours of each defect with eight neighborhood method to get the contour pixels in a certain sequence.Finally, based on the sequential contour pixels, the perimeter and the area of each defect could be estimated respectively with Euclidean distances formula and Green's theorem. The experimental result shows that this method is feasible to estimate defect size quantitatively, and digital detail enhancement technology could improve the defect detectability of pulsed thermographic system in a certain extent.
Pulsed thermographic image has the disadvantages of low-contrast, fuzzy-edge, and non-uniformity of illumination for the defect detection, thus, a defect determination method which combines digital detail enhancement(DDE) technology with maximum entropy multi-threshold segmentation method was proposed for the improvement of pulsed thermographic image. Firstly, the contrast between defects and the background was improved significantly after the image was processed with digital detail enhancement algorithm optimized with adaptive contrast enhancement(ACE) algorithm, and reducing the influence of illuminative non-uniformity on defect recognition. Secondly, the target defects with maximum entropy multi-threshold segmentation method optimized with genetic algorithm, and the contours of each defect with eight neighborhood method to get the contour pixels in a certain sequence.Finally, based on the sequential contour pixels, the perimeter and the area of each defect could be estimated respectively with Euclidean distances formula and Green's theorem. The experimental result shows that this method is feasible to estimate defect size quantitatively, and digital detail enhancement technology could improve the defect detectability of pulsed thermographic system in a certain extent.
引文
[1] Liu T ao, Li Yongfeng, Huang Wei. Application of BP neural network to quantitative identification in thermal wave NDT[J]. Infrared and Laser Engineering, 2012,41(9):2304-2310.(in Chinese)刘涛,李永峰,黄威. BP神经网络在红外热波无损检测定量识别中的应用[J].红外与激光工程, 2012, 41(9):2304-2310.
[2] Liu J, Tang Q, Yang W, et al. Defects′geometric feature recognition based on infrared image edge detection[J]. Infrared Physics&Technology, 2014, 67(4):387-390.
[3] Song Yuanjia, Zhang Wei, Wang Dongdong, et al.Defect quantitative identification and image reconstruction based on thermal wave inspection[J].Journal of Materials Engineering, 2012, 2(5):39-46.(in Chinese)宋远佳,张炜,王冬冬,等.热波检测缺陷定量识别与图像重建研究[J].材料工程, 2012, 2(5):39-46.
[4] Branchitta F, Diani M, Romagnoli M. New technique for the visualization of high dynamic range infrared images[J]. Optical Engineering, 2009, 48(9):6401.
[5] Kapur J N, Sahoo P K, Wong A K C. A new method for gray-level picture thresholding using the entropy of the histogram[J]. Computer Vision Graphics&Image Processing, 1985, 29(3):273-285.
[6] Jin Minglei, Fan Yongjie, Jin Weiqi, et al. Adaptive smoothing and dynamic range partitioning compression algorithm for IRFPA and hardware implementation[J].Optical Technique, 2013, 39(2):163-167.(in Chinese)金明磊,范永杰,金伟其,等. IRFPA的自适应细节增强算法及其硬件实现[J].光学技术, 2013, 39(2):163-167.
[7] Zhang A F, Jiang D. Adaptive contrast enhancement by a Gaussian function of local standard deviation[J].Image Matching&Analysis, 2001:7-11.
[8] Wang Xiaohong, Xu Wenbo. Study on maximum entropy multilevel threshold segmentation based on genetic algorithm[J]. Journal of Guizhou University,2007, 24(4):401-403.(in Chinese)汪筱红,须文波.遗传算法在最大熵多阈值分割的应用研究[J].贵州大学学报, 2007, 24(4):401-403.
[9] Feng Yanming. The special-shape parts geometric dimension shape parts geometric dimension measurement and defect detection based on image processing technology[D]. Hengyang:University of South China, 2015.(in Chinese)冯彦铭.基于图像处理技术的异形件几何尺寸测量及缺陷检测[D].衡阳:南华大学, 2015.
[10] Song X, Pogue B W, Jiang S, et al. Automated region detection based on the contrast-to-noise ratio in nearinfrared tomography[J]. Applied Optics, 2004, 43(5):1053-1062.
[2] Liu J, Tang Q, Yang W, et al. Defects′geometric feature recognition based on infrared image edge detection[J]. Infrared Physics&Technology, 2014, 67(4):387-390.
[3] Song Yuanjia, Zhang Wei, Wang Dongdong, et al.Defect quantitative identification and image reconstruction based on thermal wave inspection[J].Journal of Materials Engineering, 2012, 2(5):39-46.(in Chinese)宋远佳,张炜,王冬冬,等.热波检测缺陷定量识别与图像重建研究[J].材料工程, 2012, 2(5):39-46.
[4] Branchitta F, Diani M, Romagnoli M. New technique for the visualization of high dynamic range infrared images[J]. Optical Engineering, 2009, 48(9):6401.
[5] Kapur J N, Sahoo P K, Wong A K C. A new method for gray-level picture thresholding using the entropy of the histogram[J]. Computer Vision Graphics&Image Processing, 1985, 29(3):273-285.
[6] Jin Minglei, Fan Yongjie, Jin Weiqi, et al. Adaptive smoothing and dynamic range partitioning compression algorithm for IRFPA and hardware implementation[J].Optical Technique, 2013, 39(2):163-167.(in Chinese)金明磊,范永杰,金伟其,等. IRFPA的自适应细节增强算法及其硬件实现[J].光学技术, 2013, 39(2):163-167.
[7] Zhang A F, Jiang D. Adaptive contrast enhancement by a Gaussian function of local standard deviation[J].Image Matching&Analysis, 2001:7-11.
[8] Wang Xiaohong, Xu Wenbo. Study on maximum entropy multilevel threshold segmentation based on genetic algorithm[J]. Journal of Guizhou University,2007, 24(4):401-403.(in Chinese)汪筱红,须文波.遗传算法在最大熵多阈值分割的应用研究[J].贵州大学学报, 2007, 24(4):401-403.
[9] Feng Yanming. The special-shape parts geometric dimension shape parts geometric dimension measurement and defect detection based on image processing technology[D]. Hengyang:University of South China, 2015.(in Chinese)冯彦铭.基于图像处理技术的异形件几何尺寸测量及缺陷检测[D].衡阳:南华大学, 2015.
[10] Song X, Pogue B W, Jiang S, et al. Automated region detection based on the contrast-to-noise ratio in nearinfrared tomography[J]. Applied Optics, 2004, 43(5):1053-1062.