【作者】 刘艳华；

【导师】 程耀瑜；

【作者基本信息】 中北大学 ， 通信与信息系统， 2010， 硕士

【摘要】 随着高分辨率、高性能数字视频相机的成本下降,所在课题组利用高分辨率CCD数字相机和新一代影像增强器组成了高分辨率的X射线实时成像系统,和传统的图像增强器及普通模拟视频CCD相机组成的系统相比,具有空间分辨率高、抗干扰性好等特点。本论文主要工作是开发其检测软件,它具有图像获取及多种处理的功能,而且能够实现对特征参数的计算,并辅助评判人员实现对缺陷级别的判定。首先,介绍了课题组搭建的一套成像系统的硬件组成。在熟悉CCD数字相机及采集卡以后,编制了图像采集及保存软件,然后根据常规X射线检测的需要和汽车铝轮毂检测项目的实际需要,开发了一整套软件。在图像预处理部分,通过分析图像噪声,分别采用多帧叠加和中值滤波方法,达到了预期的结果。对于对比度不高的问题,采用灰度变换、直方图均衡等方法,对图像进行了增强；针对轮毂图像的特点及常规增强方法的不足,提出了一种基于Curve let变换的增强算法,提高了原图像的对比度,达到了细节区域较好的视觉效果。在图像分割阶段,采用减影技术进行分割,达到很好的分割效果,但是在分割过程中,容易造成缺陷的失真。然后在分析传统边缘提取算法的基础上,提出了基于SUSAN算子的边缘检测和基于目标区域相结合的算法,取得了理想的效果。结合企业对检测的要求及铸件缺陷评判标准,介绍了缺陷的分类、特征表达和表述方法,对周长、面积以及非几何特征进行测量,并设计了一种特征测量的实时显示系统,使缺陷特征能够实时显示出来。另外,还设计了扫描功能,显示某行或列的灰度值,实现了对某些图像缺陷的动态检测。更多还原

【Abstract】 With the cost of high-resolution, high-performance digital video camera coming down, an x-ray real-time imaging system was comprised by the group based on high-resolution digital camera and the new generation of image intensifier, which have the advantages of high spatial resolution and better anti-interference, compared with the system consisting of traditional image intensifier and ordinary analog video CCD camera. The paper was mainly to develop detection software, which contained image acquisition and processing functions, it can realize the calculation of characteristic parameters, but also be able to assist evaluation staff to determine the level of defects.First of all, the hardware structure of the imaging system was introduced in the paper. After being familiar with the CCD digital camera and acquisition card, the image acquisition and preservation software was compiled, and then according to the request of x-ray detection and the actual needs of auto aluminum wheels detection, a suite of software was developed. In the image pre-processing part, by analyzing the image noise, the multi-frame superposition and median filtering method were respectively used to achieve the expected results. For the low contrast, gray-scale transformation, histogram equalization and other methods were adopted to enhance the image. Aiming at the characteristics of x-ray aluminum wheels image and the shortage of conventional image enhancement features, a new enhancement algorithm based on Curve let transform was presented, then the contrast and the quality of the original images were improved, and the better visual effects of the minutia area was achieved. In the image segmentation stage, the subtraction technique was introduced, and got a nice result, but in the segmentation process, it was likely to cause distortion defects. Then based on the analysis of the traditional edge detection algorithm, a combination algorithm of edge detection algorithm based on the SUSAN operator and target area was proposed, and achieved the desired result. Finally, on the basis of the testing requirement and casting defects level standard, defect classification, characteristics of expression and presentation methods were introduced, and then the perimeter, area and non-geometrical characteristics were measured. Furthermore, a real-time display system was designed, so that defect features were real-time displayed. In addition, the scanning function was also designed and a row or column of the gray value can be displayed, achieving the dynamic test to a certain image defect.更多还原