Please use this identifier to cite or link to this item:
Authors: Rathod, Vijay Rajaram
Issue Date: 2011
Abstract: Over the years, the subject of welding has been productively recognized as an important aspect in the design and manufacturing of the products in the industry. It is the most common way of permanently joining two metal plates. In this process, heat is applied to the metal pieces to be joined; the metal pieces melt in resultant and fuse togetherto form a permanent bond. Common weld defects include lack of fusion, lack of penetration or excess penetration, gas cavities, slag inclusions, cracks, undercuts, lamellar tearing, and shrinking cavities etc. Regardless of the sources of error, detection of discontinuities (flaws /defects) is critical. To detect the above defects, the non-destructive testing (NDT) or non-destructive evaluation (NDE) of materials has been an area of continued growth for over sixty years. The need for NDT has increased dramatically in recent years for various reasons such as to retain product safety, in-line diagnostics, quality control, health monitoring, and security testing etc. There are at least two dozen NDT methods in use. It is only since the last decade of years that efficient computers have been used to process the images digitally in the equipment to some extent and thus helping the NDT experts in their decision. Many methods have been developed for defect detection, feature extraction and classification; however, the processing of the radio graphical weld images is a still difficult task because of presence of large number of discontinuities and other artifacts. The analysis of radio graphical images is also being helpful in categorization of weld defects. The manual interpretation of radio graphical images depends upon the level of expertise of the specialist and usually time consuming process and subjective. To remove the subjectivity in the evaluation and expediate the inspection process there is always need of automation. Most of the NDT techniques have been made automated nowadays but the interpretation of the defects is not fully standardized. With the help of image processing useful information of defects, feature extraction and classification can also be obtained which may lead to standardized inspection of weld defects. The main objective of the present research work has been to detect the flaws in weldments and develop feature extraction and classification methods. For this purpose, firstly the segmentation of radio graphical images has been aimed and implemented. Further to improve the performance of segmentation algorithms, few pre-processing algorithms especially suited for weld images were developed. xv The histogram equalization method has been applied for improving the contrast of the image. This has been followed by noise removal. For noise removal, median filters and adaptive filters were used, which are very effective in reducing the impulsive noise and enhancing the contrast of the image. The segmentation of the images forms the second stage. In present work three segmentation techniques have been implemented. The implemented segmentation techniques are: edge-based, region growing and watershed segmentation. These methodologies are compared and concluded to be effective for detection of all possible nine types of weld flaws detection i.e.(slag inclusion, worm hole , porosity, incomplete penetration, under cuts, cracks, lack of fusion, weaving fault and slag line), and there after feature extraction and classification is carried out. For implementing edge-based segmentation, edge detection algorithm best suited has been applied to determine the edges in the NDE X-ray images of weldments. To select best suitable edge detection technique; various edge detection techniques were applied on variety of radiographical weld images. A comparative study of performance of these techniques on these images was performed which emphasizes that Canny operator and second order derivatives produce better results on most of these images. If border of region is not identifiable the border is traced using different morphological functions like filling holes, clear border, dilation, erosion, closing and opening with edge detectors and border-tracing technique. By all these techniques a close contour is found which further gives an idea about the shape, size and area of a particular flaw. The other segmentation techniques implemented for segmenting the images is region growing. For implementing, region-growing approach to segment X-ray weldment images, the selection of seed value is very important aspect to start the process. The seed/seeds value is obtained with the help of histogram plot of the image showing major valleys. The seed pixel grows into the region, based on the pre-defined homogeneity criteria of gray level range in present case. This results in tracing of close boundary of the flaw. The third segmentation technique implemented is watershed transformation. The basic concept of watershed is based on visualizing an image in three dimensions i.e. two spatial coordinates versus gray levels. In such a topographic interpretation, three types of points are considered, such as, points belonging to a regional minimum, points at which a drop of water, if placed at the location of any of those points, would fall with certainty to a xvi single minimum; and points at which water would be equally likely to fall to more than one such minimum. To overcome the problem generated by watershed segmentation like poor detection of thin structures ,poor detection of significant areas with low contrast boundaries, sensitivity to noise and over segmentation multistage watershed transformation has been developed and implemented in this work. The developed segmentation algorithms have been applied and tested on the 80 radiographic images obtained from EURECTEST, International Scientific Association Brussels, Belgium, 76 actual weldments from Central Foundry Forge Plant (CFFP) of Bharat Heavy Electrical Ltd. (BHEL) India and 24 radiographs of ship weld provided by Technic Control Co. (Poland) obtained from Ioannis Valavanis Greece, (i.e. 180 X-ray weld images containing more than 200 flaws of standard defect radiographic images of welds are tested and validated.). When all these three segmentation techniques are compared, based on results produced by them the observation is obtained as following: • The edge-based image segmentation technique does not produce the closed contour. But the flaws present in the segmented images have been categorized successfully. The edge based segmentation technique works successfully on few types of flaws like slag inclusion, incomplete penetration and transverse cracks. • The region growing approach provides close contours and produces best results for lack of root penetration, undercuts and gas cavities. • Watershed segmentation techniques is best suited for identifying flaws of worm hole type, gas cavities, lack of fusion, slag inclusion and slag line. These standard image segmentation algorithms were also applied and tested on the practical database of actual weldments obtained from NDT Department of Bharat Heavy Electrical Ltd. (BHEL) Haridwar for verification and suitability of the developed algorithms. The types of flaws detected here are, like lack of root penetration, slag inclusion, gas cavities/ porosity and cracks. The approximate size and location of these flaws were physically determined by using air arc gouging and then implementing dye penetrant testing. A comparison between physically measured flaw information and that obtained from image processing algorithm was also performed. The results produced by the image processing algorithms have good correlation with the physically measured parameters. xvn Finally, a novel approach for the detection and classification of flaws in weld images using counter propagation neural network (CPNN) is presented. The method has been applied for detecting and discriminating flaws in the weld that may lead to false alarms for all aforementioned possible nine types of weld defects. The input feature vectors used are texture features and geometrical feature for training and testing purpose. According to experimental testing, 84.97% of defect samples can be classified using texture features and 94.59% using geometrical features. In summary, the present work contributes in the area of NDTweld image analysis ranging its horizons in the detection of flaws and automatic segmentation of weld images. The results obtained by the suggested methods may be useful for NDT inspectors, radiologist, or experts who may use it for welding application and treatment planning of weldments.
Other Identifiers: Ph.D
Appears in Collections:DOCTORAL THESES (Electrical Engg)

Files in This Item:
File Description SizeFormat 

Items in DSpace are protected by copyright, with all rights reserved, unless otherwise indicated.