当前位置:首页 >> 药学 >>

Nondestructive evaluation of volumetric shrinkage of compacted


Engineering Geology 85 (2006) 239 – 250 www.elsevier.com/locate/enggeo

Nondestructive evaluation of volumetric shrinkage of compacted mixtures using digital image analysis
Ali Hakan ?ren a , Okan ?nal a , Gürkan ?zden a , Abidin Kaya b,?
a

Dokuz Eylul University, Civil Engineering Department, Kaynaklar Yerleskesi, Buca, Izmir, Turkey b URS Corporation, 615 Piikoi Street, 9th Floor, Honolulu, HI 9681, United States

Received 15 November 2005; received in revised form 6 February 2006; accepted 24 February 2006 Available online 24 April 2006

Abstract Adapting image processing technology to engineering disciplines can be useful in the evaluation of the mechanical behavior of materials. Not only characteristics of granular materials, but also particulate levels of colloids can be studied using image analysis. Attempts to identify the volume change of soils or compacted specimens have been made since late 1990s. Some of the previous studies were related with the determination of the deformation field during triaxial tests while the rest were directly related with the measurement of the volumetric shrinkage strains of expansive soils. Unlike with other studies, considered volumetric shrinkage strain levels in this study were limited to 6%. The strain levels were limited, because it was noted that the maximum allowable volumetric shrinkage strain levels were 5% for evaluating the hydraulic behavior of compacted soils. The principal purpose of this study is to show the ability of image processing techniques on the quantification of the volumetric shrinkage of the compacted soils even in the small strain levels. For this purpose, a special test setup was established and a computer algorithm was developed to identify volume of the specimens from digitized images. Initially, volume changes of compacted bentonite–zeolite mixtures at various bentonite contents were measured by means of vernier caliper. Comparison of the digital measurement results with those of the manual readings showed that they were in good agreement. It appears that the proposed methodology would provide nondestructive, stable and repeatable volume measurements and is a promising approach for the quantification of volumetric shrinkage strains of compacted bentonite–zeolite mixtures even at small strain levels. ? 2006 Elsevier B.V. All rights reserved.
Keywords: Image processing; Volumetric shrinkage; Nondestructive testing; Compaction; Hydraulic conductivity; Bentonite; Zeolite

1. Introduction Desiccation cracking in soil liners and caps is a serious problem causing the initially impervious liner to act as a permeable barrier. After wetting and drying cycles upon the
? Corresponding author. E-mail addresses: ali.oren@deu.edu.tr (A.H. ?ren), okan.onal@deu.edu.tr (O. ?nal), gurkan.ozden@deu.edu.tr (G. ?zden), Abidin_Kaya@urscorp.com (A. Kaya). 0013-7952/$ - see front matter ? 2006 Elsevier B.V. All rights reserved. doi:10.1016/j.enggeo.2006.02.008

seasonal changes and fluctuations in the groundwater level, the compacted clay liner tends to either swell or shrink. The swelling/shrinkage process, however, is not fully reversible. As a result of plastic strains, many fissures and cracks develop during wetting and drying cycles. The plasticity of soils is one of the key factors that affect both swelling/shrinkage potential and hydraulic conductivity. In general, an increase in the plasticity and molding water content causes an increase in the amount of volumetric shrinkage strain of compacted soils (Kleppe

240

A.H. ?ren et al. / Engineering Geology 85 (2006) 239–250

and Olson, 1985; Phifer et al., 1994; Albrecht and Benson, 2001). In general, soils with low hydraulic conductivity are susceptible to high volumetric shrinkage. For example, Kleppe and Olson (1985) stated that compacted soils having more than 5% volumetric shrinkage strain had a high potential to exhibit major or moderate cracks and cannot be recommended as landfill liner material. Thus, researchers have suggested bentonite–sand mixtures as alternative to clayey soils to reduce the volumetric shrinkage occurring during drying/wetting cycle (Kleppe and Olson, 1985; Tay et al., 2001). In other words, shrinkage potential of a soil dictates its use as a clay liner. However, determination of shrinkage of soils is often cumbersome. Generally, the vernier caliper is used to measure the volumetric shrinkage of compacted specimens. However, it brings some discrepancies; that is, the method only considers the homogenized shrinkage of soils and does not measure whenever the soil had some discontinuity along the sidewall or at the top. The same problem also occurs during triaxial testing. In relation to triaxial tests, volume changes can be measured manually using burettes or LVDTs connected to the test cell or the specimen. Thus, the volumetric shrinkage strains directly measured by the conventional methods may be slightly lower or higher than its original shrunk size. For this reason, researchers have been interested to identify the shrinkage amount by means of digital image processing methods. The efforts were spent mostly for determining the volumetric strain of soils when tested in a triaxial cell. For example, Macari et al. (1997) used video images to compute the rate of volume change of sand during drained conventional triaxial compression test. The experimental results were compared to the volumetric strains obtained by analyzing the front view and side view images. Two video cameras were placed orthogonally to each other to take front and side view images. It was mentioned that the results were in accordance with each other unless irregular shapes occurred. In addition to this, Alshibli and AlHamdan (2001) estimated the volume change of triaxial soil specimens by locating three cameras at equal distance to capture the entire body of the specimen. The predicted volumetric strain results were in good agreement with the nominal volumetric strain measurements. Utilization of image processing technique to determine the volumetric shrinkage strains were adapted on expansive soils as well (Puppala et al., 2004). Puppala et al. (2004) showed that the manual measurements using the vernier caliper had similar volumetric shrinkage strain values with those determined by the image processing method. Although the study involves a rather wide range of volume changes on account of expansive soils (i.e. 0–40%),

differences between volumetric shrinkage strain values obtained by image processing and conventional methods happened to be in the range of 0.3% to 4.5% and were accepted as negligible in their study. However, this range of difference in two methods should be reduced when the maximum allowable limit of volumetric shrinkage strain is 5%. Thus, the objectives of this study are to determine the volumetric shrinkage performance of compacted bentonite–zeolite mixtures using image-processing technique and to compare the obtained digital measurement data with those of the conventional method. Many of the tested compacted specimens had volumetric shrinkage strain values up to 5%, which is the maximum allowable limit for landfill liner applications. Therefore, the study was focused on the limited strain level. Another limitation of the study performed by Puppala et al. (2004) is being operator dependent. For example, in that study image processing requires Scion image software. In this software both the side and top view images of the sample are processed. Such process is open to operator related errors. This is because this technique requires patching of four side view together and use of mouse to determine the dimensions of the samples. This technique is problematic for samples with low shrinkage strain since a small error made during measurement of dimensions may cause significant error in computed shrinkage strains. Thus, there is need for better image processing for soils with relatively small shrinkage strain which is the theme of this study. We used bentonite embedded zeolite mixture in our study for three reasons: (1) researchers have suggested that use of the bentonite–zeolite mixture is superior to the bentonite–sand mixture as landfill liner (Kayabal?, 1997; Kaya and Durukan, 2004; ?ren et al., 2004), (2) zeolite is known as a molecular sieve and a good adsorbent against some toxic elements and heavy metals (?ren and Kaya, in press); and (3) bentonite–zeolite mixtures have generally low volumetric shrinkage strain being less than 5%. 2. Materials and methods 2.1. Materials The test specimens were prepared by compacting zeolite and bentonite mixtures. Zeolites are hydrated aluminosilicates of alkali and alkaline earth elements with honeycomb structure. Negatively charged surface and cavity structure make zeolite as an alternative material to sand (Kayabal?, 1997; Kaya and Durukan, 2004). Whereas, bentonite, a montmorillonitic clay, is a 2:1 mineral with one octahedral sheet sandwiched between two silica sheets (Mitchell, 1993). It is an important constituent in mixtures where the

A.H. ?ren et al. / Engineering Geology 85 (2006) 239–250 Table 2 Physical and mineralogical properties of bentonite and zeolite Main minerals Zeolite Clinoptilolite Sand size fraction, (%) Silt size fraction, (%) Clay size fraction, (%) Specific gravity, Gs Liquid limit, wL (%) Plastic limit, wP (%) 95 5 – 2.28 – – Bentonite

241

achievement of lower hydraulic conductivity is needed. The self-usage of bentonite as a liner material is not suggested due to its high swelling and shrinkage potential. The chemical and physical characteristics of the zeolite and bentonite are given in Tables 1 and 2, respectively. 2.2. Methods Bentonite fractions by weight in the total weight of the mixture were set as 10%, 20%, and 30%. A total of sixteen specimens were tested in this study. The mixtures of 10% and 30% bentonite fractions had three specimens for each. The remaining ten specimens were compacted at a bentonite ratio of 20%. The target water contents were decided based on the compaction characteristics of the samples. The mixtures were blended in their air-dry state and then, proper amount of water was added to the mixture with a spray bottle. Then, the wet samples were kept in a sealed plastic bag. After a curing period of one day, the mixtures were compacted at either dry or wet side of the optimum water content. Mixtures were compacted in a mold of 101.6 mm in diameter and 116.4 mm in height. Standard (D 698) and modified Proctor (D 1557) compactive efforts were applied as specified in the ASTM. Top surface of each compacted specimen was flattened with a straightedge following the compaction procedure. Hydraulic jack was used to extrude the specimen from the compaction mold. Soon after that, compacted specimen was left for air-drying. The compaction types and conditions with respect to the maximum dry unit weights and optimum moisture contents of the samples are listed in Table 3. The volume changes at the end of the drying process were initially determined by means of the vernier caliper method. The measurements were made at several locations on the specimen. The volumetric shrinkage strain of the compacted specimens was then found considering the average diameter and height using the following equation:   Vi ?Vf ev ?%? ? ? 100 ?1? Vi where, εv is the volumetric shrinkage strain (%); Vi and Vf are initial and final volumes of the specimen (cm3),
Table 1 Chemical compositions of bentonite and zeolite Materials SiO2 (%) Zeolite 63.74 Bentonite 47.28 Al2O3 (%) 11.80 10.97 Fe2O3 (%) 1.66 1.28 MgO (%) 1.18 6.81 CaO (%) 1.86 7.90 Na2O (%) 0.51 2.81 K2O (%) 2.46 0.22

Montmorillonite – 20 80 2.76 244.4 49.4

respectively. The initial volume of the specimen was taken as the mold volume. Thus, only the final volumes of the specimens were recorded during the testing program. Extreme care was spent while measuring the heights and diameters of the specimens at several locations using the vernier caliper in order not to disturb the sample since even small spallings and grain decompositions could advarsely affect the comparison of the manual measurement data with those of the digital measurements. 3. Setup for image acquisition The general view of the test setup is shown in Fig. 1. Two planes, perpendicular to each other, were covered with a scaled paper and fixed together to form the base and the background planes for the image acquisition setup. Whenever taking a picture, a black cotton cloth was placed on the background to obtain better contrast between the specimen and the background pixels (Fig. 1). A base plate of 26 × 26 cm was fixed on the pedestal. Each specimen was placed on another plate at equal dimensions with the base plate. The midpoints of the lateral sides of both the base and the specimen plates were marked with white color so that they are lined up prior to image acquisition (Fig. 2). Exact central position of the specimen with respect to the camera was checked using a scaled paper placed on the background as a reference plane as shown in Fig. 3. The tests were performed in a dark room. Three fluorescent light sources were positioned on the base plane to make the borders of the compacted specimens more visible for digital measurements. The flashlight of the camera was also utilized during image acquisition. The camera was attached on a footing and the rotation and height of the camera were so adjusted that i) the specimen was centered and ii) the background plane was perpendicular with respect to the camera. The photographs were taken with HP PhotoSmart 945 camera of which was one of the top of the line digital camera in 2003 and it has an ability to capture 5 million pixels with an 8× optical zoom equivalent to a focal length of 37–300 mm.

242

A.H. ?ren et al. / Engineering Geology 85 (2006) 239–250

Table 3 Compaction characteristics of the bentonite–zeolite mixtures and error in measurement techniques Sample Compaction Bentonite type ratio by weight (%) 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 Std. Proc. Std. Proc. Mod. Proc. Std. Proc. Std. Proc. Std. Proc. Std. Proc. Std. Proc. Mod. Proc. Mod. Proc. Mod. Proc. Mod. Proc. Mod. Proc. Std. Proc. Std. Proc. Std. Proc. 10 10 10 20 20 20 20 20 20 20 20 20 20 30 30 30 Dry unit weight (kN/m3) 10.392 10.822 12.731 10.960 10.843 10.873 10.693 11.164 12.554 12.186 11.635 12.402 12.057 9.845 9.928 10.188 Moisture content (%) 41.7 32.9 28.4 41.8 40.2 42.5 45.1 38.6 31.5 36.3 40.0 27.2 36.7 50.9 47.6 44.4 Maximum dry unit weight (kN/m3) 10.840 10.840 12.263 10.988 10.987 10.987 10.987 10.987 12.684 12.684 12.684 12.684 12.684 11.085 11.085 11.085 Optimum moisture content (%) 39.8 39.8 32.1 38.6 38.6 38.6 38.6 38.6 24.2 24.2 24.2 24.2 24.2 32.0 32.0 32.0 Digital Manual Error in measurements measurements measurements (%) (%) (digital–manual) (%) 0.17 0.52 ? 0.21 1.88 1.88 2.28 4.24 4.26 1.33 3.08 5.86 ? 0.70 1.86 3.33 1.98 0.5 0.29 0.80 ? 0.04 1.96 1.78 2.30 4.00 3.70 1.16 2.69 5.20 ? 0.36 1.49 3.14 2.14 0.94 ?0.12 ?0.28 ?0.17 ?0.08 0.10 ?0.02 0.24 0.56 0.17 0.39 0.66 ?0.34 0.37 0.19 ?0.16 ?0.44

The aperture and shutter speed combination of the camera were set as F.8 and 1 /250, respectively in order to obtain the best results in terms of the contrast and the depth of field. Since utilization of auto focus option of the camera may result some discrepancies between two images even in the same conditions, the manual focus option was preferred to get the same focus adjustment for each image. It should be noted that even small differences in the positioning of the specimens might generate some errors while analyzing the images. For example, tiny movement of the specimen toward the camera, the volume becomes larger although the drying process was completed. One possible source of error may be attributed to small dislocations of the specimen on the plate due to sliding while rotating the specimen plate. Another source of error may originate from the camera itself while it is reactivated. On account of long duration of the test, shutting down and reactivation of the camera is unavoidable. In such conditions, small dislocations could exist in the camera position due to touching. Therefore, the center position of the object was controlled each time prior to imaging. In order to compensate for such error sources, an averaging procedure was followed during each volume measurement. In this procedure the camera was first activated and the specimen was centered with respect to the background scaled paper prior to imaging. Then, the pictures were captured from four sides by rotating the top plate. Subsequently, the specimen is removed from the plate after which it is relocated in a manner described above. Once centering is completed four images are taken for each 90° rotation of the specimen. Furthermore, the

camera is shutdown and reactivated two more times without removing the specimen from the plate and the specimen is imaged in the same manner providing 16 images for a particular measurement. Following the transfer of the images to a computer for analysis, the volume of the specimen was calculated using the average diameter and height obtained after analyzing images of each measurement. The volume thus calculated using the average height and the diameter was utilized in volumetric shrinkage strain evaluations. 4. Calculation of the average dimensions of the specimen from digital images A calibration object, made of polycarbonate material, was manufactured at nearly the same dimensions of the compaction mold so that dimension calculations could be

Fig. 1. General view of the test setup.

A.H. ?ren et al. / Engineering Geology 85 (2006) 239–250

243

Fig. 2. Marked plates and their positions prior to imaging.

made from the digital images. The picture of the calibration object was taken each time the camera was activated. Since its dimensions are already known, the established program gives two pixel coefficients for diameter and height of the calibration object. This object was used to perform the transformation of the pixel values into millimeters. Then, the pixel values of the compacted mixtures were analyzed with these coefficients. The diameter of the test specimen can be calculated directly from the tangent view of the image; whereas, the height was computed from the nearest point of the specimen relative to the camera. Calculation of the diameter of a test specimen, however, can be illustrated by the help of Fig. 4 and Eqs. (2)–(6). One should note that the radius of the calibration object was 50.85 mm for the example setup given in Fig. 4. The distance between the center of the specimen and the camera can be calculated using the definition of the angle between the line passing through the centerline of the test setup and the line that is tangent to the specimen side: r r sina ? Yx ? ?2? x sina Another equation may be written in order to express x in terms of α: a r ? b?x ?3? cosa x Eqs. (2) and (3) are combined so that Eq. (5) is obtained as in the following: r ad cosa b ? sina ? ?4? r r sina p??????????????????????! ?rb ? a a2 ? b2 ?r2 a ? sin?1 ?5? a2 ? b2

The diameter of the compacted specimen can be determined from the image as below: D? jP1 ?P2 j cosa ?6?

where, D is the diameter of the specimen, |P1 ? P2| is the distance between tangent points P1 and P2, and α is defined above.

5. Computer algorithm The computer algorithm was developed in Matlab technical computing platform. The image processing operations were programmed utilizing the Image Processing Toolbox Version 4.2 of Matlab. Other calculations including pixel counting, polynomial fitting with least square method and geometrical transformations were also programmed into the algorithm for rapid calculation of the

Fig. 3. Specimen position control before imaging: (a) one side, (b) the other side.

244

A.H. ?ren et al. / Engineering Geology 85 (2006) 239–250

Fig. 4. Plane geometry of the test setup.

test results. The Matlab code used in this study is given in the Appendix A. Furthermore, the written Matlab code with examples can be found at http://www.geocities.com/ matlabcode/imageprocessing/volumetricshrinkage/index. htm. 6. Image processing The segmentation algorithm, applied in this work, is based on the difference of the intensity level of the pixels. In order to achieve such a comparison, the true color images were converted into grayscale images. By this operation, hue and saturation information of the pixels were eliminated while retaining the luminance. In order to increase the success of the segmentation algorithm, contrast-stretching operation was applied to the images for increasing the dynamic range of the gray levels so that the specimen was separated from the darker background. The background and specimen pixels were grouped into two dominant modes in the gray level histogram of the contrast adjusted images as shown in Fig. 5. One obvious way to extract the specimen from the background is to select a threshold value separating these modes in the image, f (x, y). The segmented image, g(x, y), according to the selected threshold values could be defined as  1 if f ?x; y?zT ?7? g?x; y? ? 0 if f ?x; y?bT where T is the threshold value. Pixels labeled as 1 in the segmented image, g(x, y), correspond to the compacted specimen, whereas pixels labeled as 0 correspond to the darker background. The global image threshold value T was computed accord-

ing to the method of Otsu (1979). Since this method is based on the histogram of the image, it computes a unique threshold value for each image. Hence, this method eliminates slight contrast variations between the captured images. In this procedure, two differing threshold values were set to clarify the borders of the specimen using the developed algorithm. T1 and T2 values were obtained by adding and subtracting two constant values to T, which were determined by comparing the thresholded image with the original specimen image. The first threshold value (T1) was used for the segmentation of the left, right and the top borders of the compacted specimen (Fig. 6a). However, for the bottom border, a higher threshold value (T2) was used for the correct segmentation of the specimen and the plate (Fig. 6b). The internal gaps of the specimens in the binary images were digitally filled in order to simplify segmentation of the borders (Fig. 7). The border coordinates of the compacted specimens were detected on the acquired images by locating the points where transition from white to black pixels first occurred. In order to detect the bottom border coordinates, the image segmented with respect to the threshold value T2 was used as shown in Fig. 7b. The diameter of the compacted specimen was calculated using the average pixel values of the left and right borders, along which tangent lines of Fig. 4 are passing through. The irregularities on the specimen surface were taken into consideration by analyzing the side borders in every pixel along the specimen height. Then, the obtained values were averaged for computing the diameter. Regarding the height measurement, a slightly different approach was followed so that rough surface characteristics due to the existence of zeolite grains and disturbances along the border are to be

A.H. ?ren et al. / Engineering Geology 85 (2006) 239–250

245

Fig. 5. Gray level histogram of the contrast adjusted image.

accounted. Initially, the top and bottom border coordinates were used to fit a second-degree polynomial using the least square method. This polynomial is an approximation to the real border of the compacted specimen in the average sense. Using this seconddegree polynomial, anomalies along the borders of the compacted specimen were filtered out (Fig. 8a). However, due to the insufficient illumination of both the plate and the bottom of the specimen, the fitted polynomial along this border was unable to capture the real border of the compacted specimen resulting in smaller height measurements. Therefore, an improvement has been made to these values by only considering the values of the coordinates as they steadily increase

(Fig. 8b) so that the contact line of the specimen with the base plate is better represented. The distance between the points generated by the crossing of the polynomials and the lateral axis of the specimen were used during the height measurements. 7. Imaging techniques It is worthwhile noting for certain features of the imaging since it has been observed that they considerably affect characteristics of the acquired images. The positive influence of the flashlight and the telecentricity on image characteristics should be mentioned in this respect as in the following paragraphs.

Fig. 6. Binary image thresholded with respect to (a) T1 and (b) T2.

Fig. 7. Location of border bands.

246

A.H. ?ren et al. / Engineering Geology 85 (2006) 239–250

Fig. 8. Second-degree polynomials for height measurement (a) top and (b) bottom border of the specimen.

Fig. 10. Influence of telecentricity: (a) mapping with a focal length of 37 mm (b) mapping with increased telecentricity.

7.1. Influence of the flashlight The internal flashlight of the camera was used during the tests in order to increase contrast and to make the borders of the specimen brighter in the images. Since the background was black, the compacted specimen reflected the flashlight. Thus, the effectiveness of the segmentation algorithm, depends on the difference between the intensity levels of the pixel values, was improved by increasing the contrast of the image. The flashlight was also used to illuminate shades among the zeolite grains, especially along the interface between the compacted specimen and the plate. Fig. 9 also illustrates the influence of the flashlight, where images of the same specimen acquired without the flashlight (left side) and with the flashlight (right side) are given. 7.2. Influence of the telecentricity The perspective views of the compacted specimens include the third dimension of the scene (i.e. depth), which should be avoided during digital measurements

(Fig. 10a). While mapping the three dimensional image into a two-dimensional plane, the telecentricity of the camera lenses was increased utilizing the maximum optical zooming capacity (i.e. a focal length of 300 mm) and increasing the distance between the camera and the compacted specimen. In this case the perspective center lies near infinity enabling acquisition of much planar images of the objects (Fig. 10b). 8. Results The measurement results obtained either by manual or digital measurements are given in Fig. 11. As can be noticed in Fig. 11, there is a good match between the two measurement techniques. It should be noticed that data of each measurement are within the error margin illustrating negligible influence of the error sources defined previously (see Table 3). When the specimen is extruded from the mold, it tends to swell. Measurement results showed that negative volumetric strains are sometimes unavoidable indicating that the shrinkage strain of the specimen remained less than the swelling strain

Fig. 9. Influence of the flashlight.

A.H. ?ren et al. / Engineering Geology 85 (2006) 239–250

247

of digital measurements should be more realistic than those by manual measurements. 9. Conclusions The principal purpose of this study was that the volumetric shrinkage strain of compacted soil mixtures could be obtained even in the small strain ranges (i.e. 0?5%) by means of the digital image processing technique. The applicability of this technique was tested against manual measurements on sixteen compacted bentonite–zeolite specimens. It was found that digital and manual measurement data were generally in good agreement within the volumetric shrinkage strain range of interest. The offsets between the results of two techniques increased as the volumetric shrinkage strain get close to 5% but still differences is below 0.7%. One should note that the manual vernier caliper method is not a perfect nondestructive testing procedure since the specimens may be easily disturbed especially along their edges. This type of disturbance effect becomes important if quite low strain values are of interest. The digital measurement technique is especially promising in this respect since it offers steady, repeatable and continuous data. The proposed methodology as defined here is applicable to the volume measurement of any objects. It requires utilization of only one camera and is more economic and simpler than other methods, which need multiple cameras. Although the methodology proved to be practical and providing repeatable measurements, it is still somehow operator dependent, which can be easily improved by means of an automatic rotational movement. Acknowledgments This study is supported by the Scientific and Technical Research Council of Turkey, TUBITAK (Grant No: ICTAG I-676). The authors are grateful for this funding. Appendix A %% NONDESTRUCTIVE EVALUATION OF VOLUMETRIC SHRINKAGE OF COMPACTED MIXTURES USING DIGITAL IMAGE ANALYSIS %% ?ren, A.H.; ?nal, O; ?zden, G.; Kaya, A. (2006) %% This Matlab code was written for determination of the volume of compacted soil mixtures %% The Matlab codes and example images can be found at http://www.geocities.com/matlabcode/ clear, close all, clc %Clear all variables, close all figures and clear the screen

Fig. 11. Comparison of the results obtained by the two different measuring techniques.

taking place following the extrusion of the specimen from the compaction mold. This type of soil behavior may take place even if the water content of the specimen is low. The volumetric shrinkage strain values of the digital measurements were slightly higher than those of the manual measurements beyond 3% strain level. Puppala et al. (2004) also reached a similar conclusion although they were focused on larger strain values (i.e. up to 40%). In their study, they stated that digital measurement data exhibited 0.3% to 4.5% higher volumetric shrinkage strains than those of the vernier caliper measurements. As can be seen in Table 3, the maximum differences between two measurement techniques in this study were less than 0.7% which is far below the study of Puppala et al. (2004). Although minor camera distortion at the edges of the pictures had negligible influence on the volume measurements due to the utilization of calibration object, this difference can be attributed to the difficulties in measuring the sidewalls of the specimens using the vernier caliper especially if the specimen sides are not straight, which is generally the case. This was more pronounced for cases where higher volumetric shrinkage strains took place. It is believed that the compacted specimen may not be able to shrink in a well-defined shape for such cases. However, the digital measurement technique can take this effect into the consideration during the test since continuous readings along the sidewall can be taken. Therefore, the volumetric shrinkage strain values obtained by means

248

A.H. ?ren et al. / Engineering Geology 85 (2006) 239–250

OriginalImage=imread(‘HPIM1.jpg’); %Load the picture. In this sample, files are HPIM1.jpg, HPIM2. jpg, HPIM3.jpg, HPIM4.jpg imshow(OriginalImage),title(‘OriginalImage’); %Display the original image on the screen GrayToneImage=rgb2gray(OriginalImage); %Convert the RGB picture to grayscale image figure, imshow(GrayToneImage),title(‘GrayToneImage’); %Display the grayscale image on the screen AdjustedGrayToneImage=imadjust(GrayToneImage, stretchlim(GrayToneImage), [0 1]); %Apply contrast enhancement to grayscale image figure,imshow(AdjustedGrayToneImage),title(‘AdjustedGrayToneImage’); %Display the adjusted grayscale image on the screen T1=graythresh(AdjustedGrayToneImage)? 0.2; % Compute the threshold value(T1) by OTSU method T2=graythresh(AdjustedGrayToneImage)+ 0.15; % Compute the threshold value(T2) by OTSU method AdjustedAndThresholdedGrayToneImage1=im2bw (AdjustedGrayToneImage,T1);%Threshold the adjusted grayscale image with T1 AdjustedAndThresholdedGrayToneImage2=im2bw (AdjustedGrayToneImage,T2); %Threshold the adjusted grayscale image with T2 figure, imshow(AdjustedAndThresholdedGrayToneImage1),title(‘AdjustedAndThresholdedGrayToneImage1’); %Display the thresholded image on the screen figure, imshow(AdjustedAndThresholdedGrayToneImage2), title(‘AdjustedAndThresholdedGrayToneImage2’); %Display the thresholded image on the screen FiledAdjustedAndThresholdedGrayToneImage1= imfill(AdjustedAndThresholdedGra yToneImage1,‘holes’); %Fill the internal gaps of the thresholded image FiledAdjustedAndThresholdedGrayToneImage2= imfill(AdjustedAndThresholdedGrayToneImage2,‘holes’); %Fill the internal gaps of the thresholded image figure, imshow(FiledAdjustedAndThresholdedGrayToneImage1),title(‘FilledImage1’); %Display the filled image on the screen figure, imshow(FiledAdjustedAndThresholdedGrayToneImage2),title(‘FilledImage2’); %Display the filled image on the screen %%Clear variables for more free memory clear GrayToneImage clear AdjustedGrayToneImage clear AdjustedAndThresholdedGrayToneImage1 clear AdjustedAndThresholdedGray ToneImage2

Final=FiledAdjustedAndThresholdedGrayToneImage 1; Final2=FiledAdjustedAndThresholdedGrayTone Image2; clear FiledAdjustedAndThresholdedGrayTone Image1 clear FiledAdjustedAndThresholdedGrayTone Image2 %% Compute the mean pixel coordinate value for the left side of the %% specimen (1750 pixel were processed) imshow(OriginalImage),title(‘OriginalImage’);% Display the original image on the screen hold on for i=450:2200 for j=976:?1:0 if Final(i,j)?=Final(i,j?1) x1(i?449)=j?0.5; break end end end y1=1:size(x1,2); y1=y1+449; plot(x1,y1,‘b’) StartingPixel=mean(x1); hold on v=1:2608; %% Compute the mean pixel coordinate value for the right side of the %% specimen (1750 pixel were processed) for i=450:2200; for j=976:1952 if Final(i,j)?=Final(i,j?1) x2(i?449)=j?0.5; break end end end y2=1:size(x2,2); y2=y2+449; plot(x2,y2,‘b’) StopingPixel=mean(x2); hold on v=1:2608; l=1:1952; plot(l,1304,'y') %% Calculate the distance between left and right borders of the specimen Tangent_pix=StopingPixel?StartingPixel; %% Process the upper border of the specimen (975 pixels were processed)

A.H. ?ren et al. / Engineering Geology 85 (2006) 239–250

249

order=2; y=zeros(1,976); for j=488:1464; for i=1304:?1:0 if Final(i,j)?=Final(i?1,j) y(j?487)=i?0.5; break end end end x=488:1464; plot(x,y,‘b’) pp=polyfit(x, y, order); %Fit a second degree polynomial pval=polyval(pp,x); hold on plot(x,pval, ‘w’) l=0:2607; plot(976,l,‘y’) intersection_top=polyval(pp,976); % Calculate the intersection of the curve and axis line clear x clear y clear pval %% Process the bottom border of the specimen (975 pixels were processed) y=zeros(1,976); for j=488:1464; for i=1304:2608; if Final2(i,j)?=Final2(i?1,j) % Second image was used (Final2) y(j?487)=i?0.5; break end end end x=488:1464; plot(x,y,‘r’) %% Apply a correction for the bottom border of the specimen sbt=y(1); for i=1:fix(size(y,2)/2) if y(i)Nsbt Yy(i)=y(i); Xx(i)=x(i); sbt=y(i); else Yy(i)=sbt; Xx(i)=x(i); end end sbt2=y(size(y,2));

for i=fix((size(y,2))):?1:fix((size(y,2)))/2 if y(i)Nsbt2 Yy(i)=y(i); Xx(i)=x(i); sbt2=y(i); else Yy(i)=sbt2; Xx(i)=x(i); end end plot(Xx,Yy,‘b’) pp=polyfit(Xx,Yy,order); pval=polyval(pp,Xx); hold on plot(Xx,pval, ‘w’) l=0:2607; plot(976,l,‘y’) intersection_bottom=polyval(pp,976); format long g Tangent_pix; %The distance between tangent points K_for_Height=0.056571446651212; %This cooefficient was obtained from the calibration code. Code can be found at http://www.geocities.com/ matlabcode/ K_for_Diameter=0.0586296859024753; %This cooefficient was obtained from the calibration code Height=(intersection_bottom-intersection_top)?K_for_Height/10 %Calculate the height (cm) Diameter=(Tangent_pix?K_for_Diameter/cos(2.054? pi/180))/10 %Calculate the diameter (cm) Volume=pi?Diameter^2/4?Height %Calculate the volume of specimen (cm^3)

References
Albrecht, B.A., Benson, C.H., 2001. Effect of desiccation on compacted natural clays. ASCE Journal of Geotechnical and Geoenvironmental Engineering 127 (1), 67–75. Alshibli, K.A., Al-Hamdan, M.Z., 2001. Estimating volume change of triaxial soil specimens from planar images. Computer-Aided Civil and Infrastructure Engineering 16 (6), 415–421. Kaya, A., Durukan, S., 2004. Utilization of bentonite-embedded zeolite as clay liner. Applied Clay Science 25, 83–91. Kayabal?, K., 1997. Engineering aspects of a novel landfill liner material: bentonite-amended natural zeolite. Engineering Geology 46, 105–114. Kleppe, J.H., Olson, R.E., 1985. Desiccation cracking of soil barriers. Hydraulic Barriers in Soil and Rock. STP, vol. 874. ASTM, Philadelphia, pp. 263–275. Macari, E.M., Parker, J.K., Costes, N.C., 1997. Measurement of volume changes in triaxial tests using digital imaging techniques. Geotechnical Testing Journal 20 (1), 103–109. Mitchell, J.K., 1993. Fundamentals of soil behavior. John Wiley and Sons, Inc., New York.

250

A.H. ?ren et al. / Engineering Geology 85 (2006) 239–250 Hydraulic Conductivity and Waste Contaminant Transport in Soil. STP, vol. 1142. ASTM, Philadelphia, pp. 318–334. Puppala, A.J., Balakrishna, K., Hoyos, L.R., 2004. Volumetric shrinkage strain measurements in expansive soils using digital imaging technology. Geotecnical Testing Journal 27 (1), 1–10. Tay, Y.Y., Stewart, D.I., Cousens, T.W., 2001. Shrinkage and desiccation cracking in bentonite–sand landfill liners. Engineering Geology 60, 263–274.

?ren, A.H., Kaya, A., in press. Factors affecting adsorption characteristics of Zn2+ on two natural zeolites. Journal of Hazardous Materials. ?ren, A.H., Yükselen, Y., Kaya, A., 2004. On utilization of bentonite embedded zeolite as landfill liner. In Proc. of 7th International Symposium on Environmental Geotechnology and Global Sustainable Development, Helsinki/Finland: 1300–1309. Otsu, N., 1979. A threshold selection method from gray-level histograms. IEEE Transactions on Systems, Man, and Cybernetics 9 (1), 62–66. Phifer, M.A., Drumm, E.C., Wilson, G.V., 1994. Effects of post compaction water content variation on saturated conductivity.


赞助商链接
相关文章:
更多相关标签: