摘要
为了从红外热图像中分割和识别出碳/碳构件的缺陷,提出了一种C-V模型分割和数学形态学处理相结合的红外图像处理新方法。该方法首先用小波阈值降噪法对采集的红外热图像进行去噪声处理,提高图像的信噪比。然后采用基于C-V模型分割法对红外图像进行分割,再用Canny算子提取出缺陷轮廓;最后运用数学形态学方法平滑缺陷轮廓,去除轮廓外的孤立像素点;最终求得缺陷的实际面积。实验结果表明,该方法能很好地分割出构件中的缺陷,计算误差在6%以内;能处理十分模糊的红外图像,缺陷轮廓的提取较为准确,定位精度高。
In order to segment and recognize the defects in C/C component, an infrared image processing algorithm based on C-V model and mathematical morphology is presented. First, an oiiginal image was de-noised by wavelet threshold de-noising method to improve the SNR of the image. Then ,the de-noised image was segmented based on C- V model. The contours of defects were also detected by canny operator. Finally, mathematical morphology methods were used to smooth the contours and to remove isolated pixels outside the contours. The areas of the defects were also cal- culated. The results show that the defects are well segmented with this algorithm and that the calculation errors are within 6%. The very vague infrared images can be well processed with this algorithm. The contours of defects are also well detected with high locating accuracy.
出处
《激光与红外》
CAS
CSCD
北大核心
2014年第1期30-34,共5页
Laser & Infrared
关键词
C-V模型分割
数学形态学
红外图像处理
小波阈值降噪
C-V model segmentation
mathematical morphology
infrared image processing
wavelet threshold de-noi-sing
