Radon Transformation in Identification of Linear Feature Extraction

39 downloads 5790 Views 763KB Size Report
the welding defects using signature image processing (SIP) techniques, which includes the ... A digital SLR camera was used to photograph the test ... The RAW image is then converted to uncompressed TIFF format image. This image is ...
International Radon Journal Transformation of Advances in Identification in Softwareof Engineering Linear Feature Extraction FVolume 1, Number 1, January-June 2011, pp. 27-33

F

Radon Transformation in Identification of Linear Feature Extraction N.R. Shanker Research Scholar, Anna University.

S.S. Ramakrishanan Professor, Insititute of Remote Sensing.

P.G. Kuppusamy Senior Lecturer, Rmd Engg College, Chennai.

R. Rani Hemamalini Professor, St Peter’s Universtiy, Chennai.

ABSTRACT: The work presents a new method capable of identifying the welding defects using signature image processing (SIP) techniques, which includes the radon transformation, edge detection, photogrammetry and the morphological mathematical operators. Signature Image Processing (SIP) is a technology for analyzing electrical data collected from welding processes–usually automated, robotic welding. It is one of the reliable non-destructive testing techniques for weld inspection. This process gives the quality appraisal of a scanned weld, detailing the location and the geometry of all detected flaws. Keywords: Image processing, weld inspection, radon transformation, photogrammetry, matlab.

1. INTRODUCTION The image of welding is taken using a DSLR (Double Single Lens Resolution) camera in RAW format. A digital SLR camera was used to photograph the test plate’s surface shape using a multi station convergent network. The RAW image is then converted to uncompressed TIFF format image. This image is analysis to find the defect by using the radon transformation and the MATLAB. These samples have included the simple butt joints and following the sample welding plates were used to determine 27

International Journal of Advances in Software Engineering F

F

the defects in it. This system is available to the quality control of butt joint and used to estimate the welding strength and detect the defect. Mat lab was used to interpolate the target coordinates into a regular grid and to detect the defect in the welded pieces. 2. PROCESSING THE IMAGES Initially the processing is done by matlab with image processing tools and it is required to find the defect detection and subsequent sizing and positioning. The images were stored in the camera’s raw image format to avoid any image compression effects. Adobe Photoshop was used to convert the image from a raw image to an uncompressed TIFF. Here we taking the three welded samples Figure 1, good quality of welding, Figure 4 shows the medium quality welding, Figure 7 shows bad quality welding. First the image is get into the matlab work directory and the resolution of the image is resized because it takes a lot time to process the image in normal configuration of a system. For high resolution we need a graphic user interface with high configuration system. By processing the image with the codes we can get different images for analyzing and commands are used to process the image. 3. PHOTOGRAMMETRY The fundamental principle used by photogrammetry is triangulation. By taking photographs from at least two different locations, so-called “lines of sight” can be developed from each camera to points on the object. It is about the photographs measurement. It is an indigenous technique to obtain reliable scale measurements from images. It eliminates the systematic errors efficiently and provide the most reliable solution for collecting the image information from the raw image. It is unique in terms of considering the image-forming geometry, utilizing the information between overlapping images and explicitly dealing with the dimensions. To obtain the high accuracy, reliability and automation the system is capable of, photographs must be of the highest quality. 4. RADON TRANSFORMATION It is the technique to build the relationship between the pixel and the ground co-ordinate systems. the pixel co-ordinate system has an x co-ordinate (column) and a y co-ordinate (row). The origin of the pixel co-ordinate of the system is the upper-left corner of the image having a row and a column 28

Radon Transformation in Identification of Linear Feature Extraction F

F

value of 0 and 0, respectively. Here we are using the radon transformation. The radon computes the processing of an image matrix along specified directions. A projection of a two-dimensional function f (x, y) is a set of line integrals. The radon function computes the line integrals from multiple sources along parallel paths, or beams in a certain direction. The beams are spaced 1 pixel unit apart. To represent an image, the radon function takes multiple, parallel-beam projections of the image from different angles by rotating the source around the center of the image. The following figure shows a single projection at a specified rotation anglein this we can view the image in different angles by varying the theta value. You can compute the Radon transform of an image I for the angles specified in the vector theta.

Figure 1: Original Image Good Welding.

Figure 2: Radon Transformation Output Good Welding.

29

International Journal of Advances in Software Engineering F

F

Figure 3: Inverse Radon Image Good Welding.

Figure 4: Original Image Medium Welding.

Figure 5: Radon Transformation Output Medium Welding.

30

Radon Transformation in Identification of Linear Feature Extraction F

F

Figure 6: Inverse Radon Image Medium Welding.

Figure 7: Original Image Worst Welding.

Figure 8: Radon Transformation Output Worst Welding.

31

International Journal of Advances in Software Engineering F

F

Figure 9: Inverse Radon Image.

5. RESULT By analyzing these images we can identify the defects in the pieces. In the visual interpretation of the radon transform image of the good quality welding and then the inverse transform of the same image shows the difference in the contour as shown in Figure 2 and Figure 3. The same comparison can be seen the Medium quality welding in Figure 5 and Figure 6. In the worst quality welding shows a very big breakage in the welding path as in Figure 8 and Figure 9. Successive detection and tracking of peaks and troughs within the defect image. In order to carry out the comparison a binary image has been generated manually which contains all the real flaws of the original image. Once the processed regions have been determined with the use of the transformations and the results can be compared against the ideal binary image. This method generates a binary image from the test image where each pixel is either 0 when a regular structure features of the piece or 1 when a defect is detected. 6. CONCLUSION This paper has addressed the task of positioning of defects in image processing. A image processing technique have been developed to overcome the welding defects and their associated errors, resulting in a batch weld appraisal system where defects are automatically detected and positioned. This would make the proposed system suitable for implementations in situations requiring near real-time processing and interpretation of large Welding images, and thus these techniques are expected to greatly reduce 32

Radon Transformation in Identification of Linear Feature Extraction F

F

the possibility of human and experimental error, due to loss of concentration and visual fatigue. 7. ACKNOWLEDGEMENT N.R. Shanker, Research Scholar, Annauniversity, Chennai-25, This project was funded by Aalim muhammed salegh college of engineering, iaf avadi Chennai India. REFERENCES [1] Segmentation of Welding Defects using a Robust Algorithm Miguel Carrasco1 and Domingo Mery 2. [2] “Image Processing for Accurate Sizing of Weld Defects using Ultrasonic Time-ofFlight Diffraction”, O. ZAHRAN and W. Al-NUAIMY, Department of Electrical Engineering and Electronics, University of Liverpool, Liverpool, United Kingdom

33

Suggest Documents