·Table of Contents
New Method and Software for the weldings Non-destructive TestingEng. Mircea Tîrziu, SC "Symphony" SRL
I tested PhotoMir with many kinds of black-and-white images and I modified it to be more and more useful. In the weldings radiographies field I used radiographies made in different conditions and scanned with more kind or scanner. The radiographies presented in this article are made in the nondestructive testing department of a wagons factory, scanned with various kinds of scanners and colorized with my software.
Scanning and errors
The usual scanners introduce a large amount of errors in an image. This surprising fact is caused by the transparent nature of the radiographic film and can be proved by making a scanning of any welding radiography. To explain how apear the errors, it could be necessary to give some descriptions of the scanning process.
Any image can be splitted in very litle areas (as litle as they are named dots, points or pixels) that are arranged in rows and columns as an array. Each element of this "array" has its color. If the image is a black-and-white one each pixel has its level of gray.
To digitalize an image it is used a scanner, that "read" the level of gray of each point. Usually there are used 256 levels of gray and this means that for a black point the scanner will "read" the value 0, for a bright white will "read" the value 255, and for a gray level will read values in the above limits. The image, that are viewed as an array of points, has a correspondent array in the computer memory which holds all the values "readed". So, to scan an image is equal to fill all the elements of the array stored in the computer with the values obtained with the above rule.
One important characteristic of all the scanners is named "resolution" and it describes how many points contains an linear inch. As an example, a resolution of 300 dpi (dots per inch) means that the points have a size of about 1/10 milimeters. To see more details it is required a better resolution and it is a trend to increase this parameter. So, a 9600 dpi resolution is very usually today.
To "read" the images the scanners use two methods. The first uses light rays that are reflected by the scanned picture, a white point reflecting more light that a black one. The reflected light excits an optical device (usualy a photodiode) that transforms the light in electricity and a special electronic device transforms this in a number (usually in the 0-255 range). The second method can be used only for transparent media and uses light rays that passes throught the material. As transparent is a given point, as many light are transmitted to the other side.
Almoust all the scanners thet are on the market works with reflected light.
When I tried to use different kind of scanners to digitalize a welding radiography I obtained the folowing results:
About the colorization
Even an usual digitized black-and-white radiography has 256 gray levels, a trained man's eye cann't distinguish more that 15-20 distinct levels. Due to this fact, a lot of details remains covered. To help the man to see more levels, in some scientific areas (as in the medicine), the images are colorized. By this technique the image becomes a colored one and the man's eye being much more sensible to the color as to the gray level, the image offers more informations.
Having a black-an-white image with the gray levels in the 0-255 range, to colorize them means to give to each level of gray a color, for example to the 0 gray level we could give the red color, to the 1 gray level we could assign the orange color and so forth. The 256 levels of gray will be transformed in 256 colors. There is not obligatory to use 256 distinct colors. For example, we can set the same color for more neighbour levels of gray (as 3,4 and 5) or for very different level of grays (as 0 and 200).
Any color may be viewed as a combinations of the three basic colors: red (R), green (G) and blue (B). If we use digital colors, we can usually vary each basic color level in the 0-255 range. For example, to generate a light red we will give the maximum value to the red component (255) and the minimum value to the other components (0). We will note this by red=(255,0,0). Others colors are: black=(0,0,0), yellow=(255,255,0), white=(255,255,255) and a mediun gray=(128,128,128),
There are many other methods to generate colors and I will show only one, very useful. It uses the triplet hue (H), saturation (S) and brightness (B). In a few words, the hue is the wavelenght of the color and starts from red and ends in violet, the saturation tells us how many white color is contained in the sampled color (a saturated color has no white, whereas a gray or white color is a completely unsaturated color) and the brightness range from black to the maximum intensity of the color.
We must to establish a link between the gray levels and a parameter of the colorized image. For example, if we transform the black (r) gray (r) white scale in a red (r) yellow (r) green scale, we will know that as green is a given point in the colorized image, as white it is in the original radiography. Making this transform we will build a colorized image in which we will distinguish more levels as in the unprocessed radiography.
But this is only a step because we will find that there are more levels that cann't be separated. We will need to seek other techniquesm more suitable.
One good improvement is to convert the gray scale in a hue scale, when we will be able to see more distinct colors in the image. In many cases this could be enought, but we can observe that remains more colors that cann't be disinguished.
New colorization methods. The "SuperRainbow" technique
To increase the difference between neighbour colors I varied in the same time not only the hue, but also the brightness and the saturation. I observed that the eye is more sensible to the brightness changes only in a restricted area, around 50% of the bright colors brightness. I alse understood that the man's eye sensibility to the saturation changes is good only for the colors near saturated. For these reasons, this two parameters can be used only for low ranges.
Using these restrictions, I converted the black-and-white radiographies in colored radiographies with colors ranged in the spectrum, with simultaneous changes of the brightness and of the saturation. With this method I obtained useful images in which each level of gray can be viewed as a distinct color.
I used my software, named "PhotoMir", to create this method and, due to the fact that this program allows to change easily the colors, I improved it. The new method consists in:
The advantages of this technique are:
"SuperRainbow" technique and the weldings quality testing - results
I used more weldings radiographies to test the new method. I scanned them and I processed them by using the "PhotoMir" software. In Figure 1 is presented an radiography made in a wagons factory, at its nondistructive testing department. The image was scanned wiith an Hewlett Packard scanner with a transparence adapter. As I shown above, all the little items (as the pores and the inclusions) are near hidden. The image is unusable for any conventional nondistructive test. Figure 2 shows the same image, colorized with the "SuperRainbow" technique. I delimited an interest gray area and all the gray level outside the desired interval were set to black and white. As a first results, the image shows clearly that the welding shape is not circular; it is deformed. The image is also very useful because shows clearly the whole interest interval range at a glance. In the colored image each color corresponds to a range of gray levels. Due to the expected welding tridimensional shape, it could be expected to be seen colored strips, but an inspection over the image shows more interesting items that I named "waves" and "beaks". These details show that the opacity to the X-rays varies not linear and this fact could mean that there are areas where the thicknees of the sample has local changes or there are elements (as pores or inclusions) with changed opacity. These elements could signify local stresses that could make cracks in time and so to be harmful.for the welding. In the same time, they could mean local streches of the material that can modify the elasticity upon a perpendicular direction to the strips. Even if to know if these elements show a better welding or a worse one it must made tests, it is very probable that these elements mean changes of the welding quality. I also supose that the apparition of these items can be used to test the men and machines that make solders.
|Fig 1: The scanned radiography of a circular welding||Fig 2: The above radiography colorized with the "SuperRainbow" technique||Fig 3: A radiography that shows a "porosity"|
Because the image can show clearly all the interest interval at a glance, it become posible to analize in the same image not only the welding field, but also the welded materials, that could be affected by the welding process.
I would note that I used more variants of the "SuperRainbow" technique and I obtained more different images, showing different aspects of the same structure. I concluded that it is very important to find the right variants and the used software is very suitable to make this in a short time.
In Figure 3 is shown a portion of a radiography of a linear welding in which one can show in the upper-right side a mixture of red, yellow and green large size points. Because the color of the points reflects the opacity of the welding in the mentioned place, I supose that the picture depicts a granulated structure and will be interesting to be made future researches to establish the microscopic state of the weldings that gives such images.
|© AIPnD , created by NDT.net|||Home| |Top||