MULTI-SENSOR BASED CONDITION ASSESSMENT SYSTEM FOR BURIED CONCRETE PIPE

  • : Ms Word Format
  • : 100 Pages
  • : ₦3000
  • : 1-5 Chapters
  •  
  • Click to DOWNLOAD Materials

MULTI-SENSOR BASED CONDITION ASSESSMENT SYSTEM FOR BURIED CONCRETE PIPE

Abstract

The American Society of Civil Engineers’ 2005 Report Card for America’s Infrastructure gave a D− grade to water/wastewater infrastructure . It has been estimated that up to 40% of the United States’ underground infrastructure will have failed or will be on the brink of failure within 20 years, unless efforts are initiated to renew it. But system renewal requires adequate funding. According to an April 2000 report by the Water Infrastructure Network (WIN) Agency, “America’s water and wastewater systems face an estimated funding gap of $23 billion a year between current investments in infrastructure and the investments that will be needed annually over the next 20 years to replace aging and failing pipes and meet mandates of the Clean Water Act and Safe Drinking Water Act”. This necessitates the need to monitor, detect and prevent any unforeseen failures in the working of underground pipelines that are complex in nature. A reliable pipeline assessment system is necessary so that pipeline operators can develop cost-effective maintenance, repair, and rehabilitation programs. This research proposes an automated ultrasound-immersion-based inspection system that can add complementary pipe information (depth perception) to existing surface image assessments done on concrete pipes commonly used as gravity stormwater and sewer pipes. Most municipal pipeline systems in North America are inspected visually by mobile Closed Circuit Television (CCTV) systems to access the in-situ condition of buried pipes. The video images are examined visually and classified into grades according to extent of damage against documented criteria by human operators prone to fatigue, subjectivity and ambiguity. Additionally, current imaging systems like Sewer Scanning & Evaluation Technology (SSET) and CCTV are able to provide information from within the pipe regarding surface cracks in 2-D only and do not have the capability to provide depth perception.

This thesis provides a proof-of-concept of an automated ultrasound-immersion-

 

based inspection system to detect defects in buried concrete pipes. The inspection system is proposed as a two step approach. The first step is called a reconnaissance mission that uses the ultrasound transducer to scan a region of interest. A signal interpretation and classification scheme coupled with a post processing algorithm is proposed that classifies the region of interest into a clean or defect region. If the scanned region of interest belongs to a defect, the second step characterizes the region of interest with a C-scan imaging process that provides depth perception. Results have shown that the feature extraction, classification and post processing schemes proposed in this thesis provide a sound proof-of-concept for developing this inspection system into a field applicable tool. Such a tool can aid asset managers to quickly evaluate the status of their buried infrastructure, ultimately leading to a sustainable asset management system.

Table of Contents

List of Figures                                                                                                               ix

List of Tables                                                                                                              xiii

Acknowledgments                                                                                                     xiv

Chapter 1 Introduction                                                                                                1

1.1                          Motivations . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .                             3

1.2     Objective & Scope                      . . . . . . . . . . . . . . . . . . . . . . . . . . .                         5

1.3                         Contributions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .                            5

1.4                      Thesis Organization . . . . . . . . . . . . . . . . . . . . . . . . . . .                         8

Chapter 2 Background                                                                                              13

2.1                         Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .                          13

2.2             Overview of Pipeline Assessment Techniques . . . . . . . . . . . . .             13

2.2.1        Destructive Testing Methods for Defect Assessment . . . . .        14

2.2.2     Non-Destructive Testing Methods for Defect Assessment    . .     15

2.2.2.1               Non-Visual NDT Methods . . . . . . . . . . . . . .               16

2.2.2.2      Visual NDT Methods                . . . . . . . . . . . . . . . .               17

2.3    Automatic Image-based Inspection                . . . . . . . . . . . . . . . . . .                19

2.4    Image Processing And Segmentation               . . . . . . . . . . . . . . . . .                20

2.4.1                   Image Pre-processing . . . . . . . . . . . . . . . . . . . . . .                   21

2.4.1.1     Gray Scale Transformation             . . . . . . . . . . . . .             21

2.4.1.2                  Image Smoothing . . . . . . . . . . . . . . . . . . .                 22

2.4.1.3     Color Image Processing              . . . . . . . . . . . . . . .              22

2.4.2    Image Segmentation                  . . . . . . . . . . . . . . . . . . . . . .                   26

2.4.2.1     Threshold-based Segmentation          . . . . . . . . . . .            26

 

2.4.2.2               Edge-based Segmentation . . . . . . . . . . . . . .               27

2.4.2.3              Region-based Segmentation . . . . . . . . . . . . .              28

2.4.3                  Mathematical Morphology . . . . . . . . . . . . . . . . . . .                  29

2.5                        Ultrasonic NDT . . . . . . . . . . . . . . . . . . . . . . . . . . . . .                         30

2.5.1               Basic Types of Ultrasound Waves . . . . . . . . . . . . . . .                32

2.5.2         Contact And Non-contact Ultrasound Transduction . . . . .         33

2.5.3                  Ultrasonic NDT of Concrete . . . . . . . . . . . . . . . . . .                  33

2.6                       Feature Extraction . . . . . . . . . . . . . . . . . . . . . . . . . . .                        36

2.7    Classification                        . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .                         37

2.7.1     Statistical Classification                 . . . . . . . . . . . . . . . . . . . .                  38

2.7.1.1          Prior probabilities and the Default rule . . . . . . .          39

2.7.1.2                  Separating Classes . . . . . . . . . . . . . . . . . .                 39

2.7.1.3                Misclassification Costs . . . . . . . . . . . . . . . .                40

2.7.2               Artificial Neural Networks (ANN) . . . . . . . . . . . . . . .               40

2.7.2.1                     Learning . . . . . . . . . . . . . . . . . . . . . . . .                     41

2.7.2.2     Learning Paradigms              . . . . . . . . . . . . . . . . .                43

2.7.2.3                  Features of ANNs . . . . . . . . . . . . . . . . . . .                  44

2.8    Summary                         . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .                          47

Chapter 3 CCTV Pipeline Image Pre-processing                                                  57

3.1                         Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .                          57

3.2            Method I: Non-Linear Quadratic Filtering Method . . . . . . . . . .            59

3.3            Method II: Magnification Of Dark Image Features . . . . . . . . . .             61

3.4    Summary                         . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .                          62

Chapter 4 CCTV Pipe Crack Segmentation                                                            69

4.1                         Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .                          69

4.2                         Edge Detection . . . . . . . . . . . . . . . . . . . . . . . . . . . . .                         70

4.3    Mathematical Morphology Based Operators            . . . . . . . . . . . . .             71

4.4                   Analysis of Crack-Like Pattern . . . . . . . . . . . . . . . . . . . . .                   73

4.4.1          Crack Detection Approaches and Previous work . . . . . . .          73

4.4.2     Modeling Crack Like Pattern                . . . . . . . . . . . . . . . . .                74

4.5                    Detection of Crack Features . . . . . . . . . . . . . . . . . . . . . .                     75

4.5.1     Spectrum of SSET imagery                 . . . . . . . . . . . . . . . . . .                75

4.5.2          Geometry based recognition of crack features . . . . . . . . .          75

4.5.3       Final segmentation based on curvature characteristics . . . .       76

4.6                Flow Chart of The Proposed Algorithm . . . . . . . . . . . . . . . .                77

4.7                   Generalization to Pipe Images . . . . . . . . . . . . . . . . . . . . .                    78

4.7.1     Background and Color Variation Images           . . . . . . . . . . .            79

4.7.2    Performance Evaluation                 . . . . . . . . . . . . . . . . . . . .                  79

4.7.3    Parameter Selection                   . . . . . . . . . . . . . . . . . . . . . .                   80

4.8                           Results . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .                            81

4.9 Comparison with Conventional Methods: A Qualitative and Quantitative Discussion . . . . . . . . . . . . . . . . . . . . . . . . . . . 82

4.10 Summary                         . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .                          84

Chapter 5 Ultrasonic Inspection Of Wastewater Concrete Pipe                       99

5.1                         Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .                          99

5.2                   Suitable Approaches For Concrete Pipe Inspection . . . . . . . . . . 100

5.2.1     Practical Issues Governing Selection of Inspection Method      . 101

5.2.2                          Ultrasonic Guided Wave Inspection . . . . . . . . . . . . . . 102

5.2.3                                           Impact Echo . . . . . . . . . . . . . . . . . . . . . . . . . . . 103

5.2.4                           Ultrasound Immersion Inspection . . . . . . . . . . . . . . . 105

5.2.5                             A, B and C-scan Representations . . . . . . . . . . . . . . . 106

5.3                                        Experimental Program . . . . . . . . . . . . . . . . . . . . . . . . . 107

5.3.1                                            Objectives . . . . . . . . . . . . . . . . . . . . . . . . . . . . 108

5.3.2                                   Specimen Description . . . . . . . . . . . . . . . . . . . . . . 108

5.3.3                       Experimental Setup and Data Collection . . . . . . . . . . . 109

5.4                                                   Results . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 111

5.5                                             Saturation Issues . . . . . . . . . . . . . . . . . . . . . . . . . . . . 114

5.6    Summary                                                . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 115

Chapter 6 Feature Extraction And Classification of Ultrasonic Sig-

nals                                                                                                      149

6.1                                               Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 149

6.2                                            Feature Extraction . . . . . . . . . . . . . . . . . . . . . . . . . . . 151

6.2.1       Clustered Discrete Wavelet Transform Based Feature Ex-

traction                                            . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 152

6.2.1.1                        Discrete Wavelet Transform . . . . . . . . . . . . . 152

6.2.1.2 Clustering Procedure . . . . . . . . . . . . . . . . . 155 6.2.2 Subset and Compressed Discrete Wavelet Transform Based Feature Extraction . . . . . . . . . . . . . . . . . . . . . . . 159

6.3    Classification                                             . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 160

6.3.1    Proposed Multilayer Perceptron (MLP) Neural Network Clas-

sifier . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 162

6.3.2     Statistical Classifier for Performance Comparison            . . . . . . 166

6.4                              Experimental Results and Discussion . . . . . . . . . . . . . . . . . 168

6.5                             Post Processing for Depth Perception . . . . . . . . . . . . . . . . . 169

6.5.1                                          Assumptions . . . . . . . . . . . . . . . . . . . . . . . . . . . 171

6.5.2     Proposed Post Processing Algorithm                      . . . . . . . . . . . . . 172

6.6    Summary                                                . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 174

Chapter 7 Prototype Development                                                                      192

7.1                                               Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 192

7.2                                    Prototype Inspection Design . . . . . . . . . . . . . . . . . . . . . . 193

7.3                              Experimental Setup & Data Collection . . . . . . . . . . . . . . . . 193

7.4                                                   Results . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 194

7.5                                         Field Implementation . . . . . . . . . . . . . . . . . . . . . . . . . . 195

Chapter 8 Conclusions                                                                                            204

8.1                                               Contributions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 204

8.2                                            Future Directions . . . . . . . . . . . . . . . . . . . . . . . . . . . . 206

8.3                                          Concluding Remarks . . . . . . . . . . . . . . . . . . . . . . . . . . 209

Bibliography

Chapter 1 Introduction

Beneath North America’s roads lie 1.6 million miles of pipelines that bring purified water to homes and carry away waste water (sewage and storm water). Aging wastewater management systems discharge billions of gallons of untreated sewage into U.S. surface waters each year. The Environmental Protection Agency (EPA) estimates that the nation must invest $390 billion over the next 20 years to replace existing systems and build new ones to meet increasing demands. These buried infrastructure systems have been functioning longer than their intended design life with little or no repair. They are aging and in a progressive state of deterioration. Maintenance and rehabilitation of pipeline systems pose a major challenge for most municipalities in North America given their budgetary constraints, demand on providing quality service, and the need for preserving their pipeline infrastructure. Neglecting regular maintenance and rehabilitation (M&R) of these buried pipelines adds to life-cycle costs and liabilities, and in extreme cases causes stoppage or reduction of vital services. Unsurprisingly, as shown in figure 1.1, the American Society of Civil Engineers’ 2005 Report Card for America’s Infrastructure gave a D− grade to water/wastewater infrastructure . It has been estimated that upwards to 40% of the United States’ underground infrastructure will have failed or will be on the brink of failure within 20 years, unless efforts are initiated to renew it (Technical Report (WIN), 2001, 2002). But system renewal requires adequate funding. According to an April 2000 report by the Water Infrastructure Network (WIN) Agency, “America’s water and wastewater systems face an estimated funding gap of $23 billion a year between current investments in infrastructure and the investments that will be needed annually over the next 20 years to replace aging and failing pipes and meet mandates of the Clean Water Act and Safe Drinking Water Act” (Technical Report (WIN), 2001). This necessitates the need to monitor, detect and prevent any unforeseen failures in underground pipelines.

Accurate pipeline condition assessment is vital to developing a cost effective and efficient pipeline M&R program. At present, the assessed condition of buried pipes is based on the subjective visual inspection of closed circuit television (CCTV) surveys (Gokhale et al., 1997). CCTV surveys are conducted using remotely controlled vehicle carrying a television camera through a buried pipe. The data acquired from this process consist of videotape, photographs of specific defects, and a record produced by the technician. Typical scanned image of CCTV surveys is shown in figure 1.2. It is a well known fact that the diagnosis of defects depends on experience, capability and concentration of the operator thus making the detection of defects error prone.

Large number of new technologies such as Sewer Scanner and Evaluation Technology (SSET), laser based scanning systems, etc. have made it possible to obtain high quality images of buried pipes. SSET is an innovative technology for obtaining unfolded images of the interior of buried pipes (Iseley, 1999). This is achieved by utilizing scanner and gyroscope technology. Typical scanned images of SSET surveys are shown in figure 1.3. Inspite of buried imaging technologies making giant strides in recent years, the basic means of analysis remain unchanged: a qualified technician is still required to identify defects on a television monitor. One of the objectives of this thesis will be to address the above limitation.

Additionally, a defect that appears on the surface to be insignificant (less than 5 mm mouth opening) might actually exist throughout the thickness of the pipe. An operator would usually classify such a defect as a ‘minor defect’ and shift attention to those that may appear critical on the surface but do not extend into the depth of the pipe. This may prove to be catastrophic because of the defect being classified as ‘minor’ due to lack of enough information. It may well be the case that the so called ‘minor’ defect will lead to a pipe collapse much earlier than the other cracks that were classified to be critical based on surface analysis only. Hence, any crack (or defect) classification system that primarily depends on surface characteristics is incomplete and needs to be complemented with additional depth perception to provide for a reliable, accurate and effective buried pipeline asset management system. Therefore the development of an automated buried pipeline condition assessment system that can provide additional depth perception of defects is the main objective of this research study. Main efforts are placed on investigating the use of ultrasound acoustics to acquire depth perception and complement 2-D crack features available from the SSET camera. Additionally, improving the interpretation of surface scanned images and ultrasound signals by implementing new algorithms and techniques from image processing, feature extraction and pattern classification are also within the scope of this study.

As mentioned before, the main focus of this thesis is on developing proof-ofconcept of an intelligent automated ultrasound inspection system that can inspect concrete pipes and provide depth perception. In particular, this thesis explores the use of various signal and image processing concepts, nonlinear filtering, ultrasound acoustics, feature extraction and pattern classification techniques that are combined to provide a two step inspection system. The proposed inspection system can lead to overcoming many limitations of the current manual inspection practice and can provide a more accurate assessment of buried pipe conditions.

The next sections outline motivation for this research, followed by contributions and a description of the organization of this thesis.

1.1       Motivations

Visual inspection based on closed circuit television surveys is widely used in North America to assess the condition of buried pipes (Wirahadikusumah et al., 2001). The human eye is extremely effective at recognition and classification, but it is not suitable for assessing pipe defects in thousand of mile of pipeline due to fatigue, subjectivity and cost. This drawback of the present method of visual inspection is one of the main motivations behind this research study in developing a sophisticated condition assessment system.

The present state of technology only affords a 2-D view of pipe surface characteristics using the SSET scanning method and there are justifiable reasons to know the depth (3rd dimension) information of defects (e.g., cracks). Hence, acquiring depth perception of defects that are presently visible only at the surface (in 2-D) through the current inspection technology, is of considerable interest to the buried asset management community. This additional data source has the potential to solve problems of fatigue, subjectivity, and ambiguity by the very fact that defects can now be interpreted based on their depth in pipe thickness.

Most of the literature concerning the detection/classification of defects based on imaging in civil structures deals with the analysis of pavements and concrete/steel distresses(Cheng, 1996, Cheng and Miyogim, 1998, Haas and Hendrickson, 1990, Walker and Harris, 1991). In the past few years, research on assessing the condition of buried pipes has gained recognition in the buried infrastructure asset management community through the efforts of Isley et al (1997), Isley (1999), Wirahadikusumah et al (1998), and Feiguth and Sinha (1999). In analyzing scanned buried pipe images, one needs to consider complications due to the inherent noise in the scanning process, irregularly shaped cracks, as well as the wide range of pipe background patterns. One of the major problems is detecting defects (especially cracks) that are camouflaged in the background of corroded areas, debris, patches of repair work, and areas of poorly illuminated conditions. But, most of the available methods of detection and classification are strictly based on 2-D information provided by the SSET scanned images. Approaches that can provide a complete description of the defects based on 3-D features do not exist and hence provide for an interesting research problem.

In light of the above discussion, there is sufficient motivation for the development of computationally viable, efficient and robust methods for image preprocessing, segmentation, and detection of surface defects from SSET images. However, the main focus is on acquiring depth perception through ultrasound acoustics based methods. Many researchers have been working on characterizing concrete using ultrasound acoustics. Concrete has always been a difficult material to deal with from an ultrasonic point of view owing to its heterogeneity and high attenuating properties (Malhotra and Carino, 2004). These aspects also motivated the development of an ultrasound-based inspection system that is capable of detecting defects and then characterizing them with respect to depth perception in representative concrete pipe specimens.

1.2        Objective & Scope

The primary objective of this research is to develop an automated buried concrete pipe inspection system based on scanned images obtained from SSET camera and signal/image data from the ultrasound sensor. Specifically, the scope of this thesis begins with researching new techniques to improve the detection of cracks under varying background, pipe color and complicated defect patterns from surface scanned images. This step provides accurate two dimensional data about various kinds of cracks on the internal surface of the concrete pipe. The second and most important objective of this study is to propose an ultrasound acoustics-based methodology to acquire depth perception about surface cracks outlined in the previous step. The scope of this study concludes with the implementation of an ultrasound inspection system consisting of a two step approach. At the point in time when this thesis research had commenced, the following work had been accomplished (Sinha, 2000, Fieguth and Sinha, 1999, Sinha et al., 1999, Sinha and Fieguth, 2001):

  • Automatic segmentation of buried pipe images
  • Detection of cracks in segmented pipe images
  • Feature extraction of image features for defect classification

The contributions of this thesis that builds upon the above accomplishment are listed below.

1.3       Contributions

CCTV Image Processing & Segmentation

Contrast enhancement is a pre processing step before segmentation. Contrast enhancement can be defined as a radiometric enhancement technique used to improve the visual contrast of an image. In the analysis of the objects in images, it is essential to distinguish between the objects of interest and “the rest.” This latter group is also referred to as the “background.” The techniques that are used to find the objects of interest are usually referred to as segmentation techniques and the process is known as segmentation – segmenting the foreground from background. In this thesis, two approaches for enchancing the contrast of CCTV images for better interpretation have been developed. The first approach uses a modified version of unsharp masking through non-linear quadratic filtering to enhance the crack features and suppress the background characteristics in an image. This is accomplished in the first proposed method using non-linear quadratic filtering. The second approach increases the contrast of the dark pixels from the estimated “background” image by comparing the intensity of each pixel in the original color image with that of the background image. If the difference in intensities is higher than a given threshold, it darkens the dark pixels and lightens the background pixels thus enhancing the contrast between crack and background.

To determine the efficiency of both the proposed methods, a simple, robust and efficient algorithm for detecting crack patterns in pipeline images is developed. The algorithm consists of three steps, contrast enhancement, morphological treatment and curvature evaluation in the cross direction, and finally the alternating filters that produce the final segmented binary crack map. The proposed approach can be completely automated and experimental results demonstrate that algorithm is effective for segmenting CCTV images with varying background, color, and crack patterns.

Development of Ultrasound-based Inspection System

In this thesis, various ultrasound-based inspection techniques that may be employed to acquire depth perception data for defects in buried concrete pipes are evaluated. Two methods, guided wave and impact echo, are experimentally examined for suitability with concrete inspection. Both these techniques have their own limitations in terms of their suitability for concrete pipe inspection. The guided wave technique is successful with a transducer configuration that cannot be implemented in a buried pipe inspection scenario. The frequency range of operation in concrete pipe with 60 mm thickness proves to be fairly high for typical impact-echo applications.

A significant contribution of this thesis is the development of an ultrasound immersion based technique that uses water couplant as a suitable approach for inspecting concrete pipes. Experimental results, in an amplitude time series signal (A-scan) and image format (C-scan), of several concrete samples with defects of interest to the water/wastewater pipe community show that it is possible to make interpretations about the presence of defect and provide depth perception by this method. The results prove that this methodology is successful in providing additional insight into the condition of concrete samples under investigation.

Development of Feature Extraction & Classification Schemes

In buried concrete pipe defect analysis, the main objective is to identify and classify regions of the pipe as “clean” or “defective”. Further classification of defects into various sub-classes like hole, fracture, crack and hairline crack is possible based on data provided by the proposed ultrasonic inspection system.

Classification is a statistical procedure in which individual items are placed into groups based on quantitative information on one or more characteristics inherent in the items (referred to as traits, variables, characters, etc) and based on a training set of previously labeled items.In this thesis, a multi-layer perceptron (MLP) neural network classifier that uses discriminatory features from the ultrasonic signal is developed. The proposed approach uses wavelet analysis to decompose the signal into its useful information components and then employs an unsupervised clustering scheme to extract feature vectors that represent the class of the signal. Wavelet analysis refers to the representation of a signal in terms of a finite length or fast decaying oscillating waveform (known as the mother wavelet). This waveform is scaled and translated to match the input signal. The MLP classifier classifies the signal into its appropriate class based on extracted features of interest from the wavelet filtered signal.

Development of Inspection Approach

An overall contribution of this thesis is in developing a framework for the inspection system. It is proposed that the inspection system be envisioned as a two step approach consisting of the reconnaissance mode and characterization mode. In the reconnaissance mode, the ultrasound transducer is used to scan a region of interest (ROI) and acquire A-scan signals. The ROI A-scan signals are passed through the classifier and identified as belonging to defect or clean class. If the scanned ROI belongs to a defect of interest, the second step of the approach is to characterize the ROI with a C-scan imaging process for depth perception. A relevant contribution of this thesis is in developing a post processing scheme that provides a high confidence that C-scan imaging will only be ‘triggered’ when a defect of interest is detected. The feature extraction, classification and post processing schemes proposed in this thesis provide a sound proof-of-concept for developing this inspection system into a field applicable tool.

1.4       Thesis Organization

Chapter 2 presents the background relevant for understanding the automated buried pipe inspection system. It begins with a broad overview of various pipeline assessment techniques in general. It then briefly introduces the methodology for automated image-based inspection. The next two sections discuss image segmentation and feature extraction methods. Next, we review ultrasound nondestructive testing and discuss various approaches applicable to concrete. Finally, a brief review of pattern recognitions tasks is provided at the end of this chapter.

Chapter 3 presents two approaches for enchancing the contrast of CCTV images for better interpretation. A modified version of unsharp masking which computes the enhancement map for thresholding ,so that crack features are enhanced and background is suppressed, is discussed. The chapter also discussed the second approach which increases the contrast of the dark pixels from the estimated “background” image by comparing the intensity of each pixel in the original color image with that of the background image. Next, the chapter discusses the efficiency of both these methods by developing a crack detection filter and applying an adaptive algorithm to contrast enhanced images.

In Chapter 4, a simple algorithm for detecting crack patterns in pipeline images is adapted and implemented. This chapter begins by presenting mathematical foundations of the morphological operations used in the proposed algorithm. Next, it discusses the implementation strategy, followed by performance evaluation of the proposed method with other conventional detection techniques.

Chapter 5 describes the ultrasound-based inspection techniques that may be employed to acquire depth perception data for defects in buried concrete pipes. First, it presents the challenges in developing an inspection methodology for buried concrete pipes. Next, several techniques like guided wave and impact echo are examined for suitability with concrete inspection. The chapter proposes ultrasound immersion technique as a suitable approach for inspecting concrete pipes. This chapter also presents experimental results of several concrete samples with defects of interest to the water/waster water pipe community. A-scan signal and C-scan image representations from a region of interest (ROI) are presented to make interpretations about the presence of defect and provide depth perception.

Chapter 6 describes the feature extraction methods and pattern recognition strategies for the classification of ultrasound inspection data. A formulation of the multilayer perceptron neural network with input and output parameters is presented in this chapter. This chapter also discusses a feature extraction scheme based on discrete wavelet transform and unsupervised clustering to extract signal features for classification. A post processing scheme to interpret the classifier outputs and finally classify the signals into an appropriate class taking into consideration some apriori knowledge of the problem is discussed towards the end of this chapter. Chapter 6 also provides a framework for the inspection system consisting of two steps, a reconnaissance mode and a characterization mode.

Chapter 7 briefly presents the concept, implementation, and results from a wa-

ter bubbler system that emulates the immersion environment discussed in chapters 5 and 6 in a real-world scenario on an actual concrete pipe specimen with defect.

Chapter 8 summarizes the outcome of this thesis, lists major contributions of this work, details future directions for research, and provides concluding remarks.

Figure 1.1. ASCE’s 2005 report card on America’s infrastructure. Image courtesy www.asce.org/reportcard/2005.

(a)                                                  (b)

Figure 1.2. (a) and (b) forward vision (FV) image from CCTV camera

(a)                                            (b)

(c)                                           (d)

Figure 1.3. Typical images of buried pipe scanned by Sewer Scanner and Evaluation

Technology (SSET) camera

MULTI-SENSOR BASED CONDITION ASSESSMENT SYSTEM FOR BURIED CONCRETE PIPE

Leave a Reply