Analysis of Different Crack Detection Methods

Priya Ranjan Muduli et al. (2013) have identified a crack detection method in which they have combined two of the methods and the methods were Hyperbolic Tangent filtering and Canny edge detection algorithm. They have also used Haar Discrete Wavelet Transform Algorithm (HDWT) in which the algorithm decomposed a signal in to two sub signals. The one subsignal was the average and the other was the difference.

E. Balasubramanian et al. (2015) have investigated a novel approach combining both HSV thresholding technique and hat transform for the detection of cracks and also the parameters of the cracks were obtained. The overcome the difficulties of previous researchers they have proposed the new method. In HSV the object can be extracted by properly setting the limits in the algorithm HSV. By using the simple grey scale thresholding the surface degradation assessment could be done and it has been used to classify the regions that are not affected.

Arun Mohan and Sumathi Poobal et al. (2016) have investigated various crack detection methods. In their paper they have done an analysis based on various parameters. Those analysis was based on the objectives, techniques, accuracy level, error level and datasets. They have not addressed the different classification techniques and its issues.

Rizvi Aliza Raza et al. (2017) have proposed a new method for the crack detection of railway tracks. They have installed video cameras in different areas of the track for the image acquisition and different methods were used for the crack detection and the methods were de-noising, histogram equalization, morphological operation etc.

Roberto Medina et al. (2017) have used Gabor filters to the rotation method and their method has been used to show different orientations along its length for single crack. Their method was insensitive to brightness.

Haiming Liu et al. (2016) have analyzed a method for the detection of cracks in plastic surfaces. Plastic surfaces may have line-like cracks. They have applied reduction method for the removal of noise, image gradient for the reconstruction of the crack images; to identify the crack shape based optimal method has been used in their research. Their proposed approach was better than Otsu method and clustering method.

Aliza et al. (2017) have investigated a crack detection method for aircraft or automobile applications. In the case of aircraft and automobile applications the cracks could not be detected from one image and it also requires more time. So they have used methods such as morphological operation, thresholding, Canny edge detection etc.

Mojtaba et al. (2016) have used some pre-processing techniques in which it eliminates the impact of non-uniform background and pavement markings. They have also used morphological operations to enhance the features of the cracks.

Romulo et al. (2016) have used a feature extraction method in their research. They have used the feature extraction method to differentiate the outdoor images. Their method was having a classification for segmenting the window based on color, crack detection for clustering, selection of particles for particle filtering and classification on direction with quantitative analysis for least square method.

Rabihamhaz et al. (2016) have used three methods for crack detection in their research. The three methods were Minimal Path Selection which was used for the detection of cracks, Pseudo Ground Truth(PGT) for the evaluation of cracks, DSC (DICE similarity Coefficient) were used in their proposed method Chen et al. (2016) have proposed another method known as wavelet transform and KD-tree. They have used low resolution images.

M. A. Hassan et al. (2015) have proposed contrast adjusted Otsu method for the detection of cracks. Their purpose was the defect detection in titanium coated aluminum surfaces. They have used purely histogram based algorithm.

Tung-Yen Li et al. (2013) have identified new methods called as Optical interference Pattern sensing method and neural network classification method. Their purpose was defect detection in TFT and LCDs. Their method was found to be very robust and reliable.

Xiaolong Bai et al. (2014) have introduced a method called Phase-only Fourier transform (POFT) and Template matching for the detection of cracks. Their purpose was to detect defects in electronic chips.Their method was simple and it was an easy method for defect detection.

W.L. Woo et al. (2015) have analyzed Bayesian statistics methodology and adaptive sparse representation method for defect detection in metal surfaces. Their method was more economical and gives higher detection performance in metal surfaces.

Anandhanarayanan Kamalakannan et al. (2012) have proposed a fuzzy image thresholding and linear classifier model for surface defect detection. Their purpose was to find defects in mandarin fruits. In their proposed methodology parameters has to be selected prior to the running of the algorithm and their method lacks an automated feature selection.

Deng Sier et al. (2010) have proposed Otsu’s thresholding method for defect detection in bearing surfaces. Their propose method used a purely histogram based algorithm but it fails to properly select a threshold when it is located near a local top on the histogram.

Amir Hossein Aghamohammad et al. (2011) have investigated a Particle swarm optimization algorithm for defect detection. Their purpose was the crack detection in the surfaces of solar cell panel. The method incorporates a low convergence rate within the repetitive method

Gagan Kishore Nand et al. (2014) have analyzed a new method called Entropy Segmentation for the defect detection. Their purpose was to find the fault recognition of steel surfaces. Their method successfully identifies the faults in steel surfaces like water droplet blister.

Zhou et al. (2013) have proposed a method called circular region projection histogram and an algorithm based on sparse representation. Their purpose was to detect faults in water cap surfaces. Their algorithm was effective in the defect detection of bottle caps.

A.Sada Siva Sarma et al. (2013) have proposed a technique called Texture Feature Extraction using a three level 2D Haar wavelet transform and artificial neural network classifier. Their purpose was the fault detection on the surface of hot rolled steel sheets. Their method was suitable for checking the surface defect of low resolution and non-uniform lighting images.

Anders Landstrom et al. (2012) have used techniques such as morphological image processing and statistical classification method. Their purpose was to analyze cracks in steel slabs. Accuracy of their method was low since some cracked regions were completely missed.

Linghui et al. (2011) have proposed wavelet transform C V model. Their purpose was to find the defects of 3D industrial CT images. Their proposed method was not appropriate for fast applications and also their method could not achieve good segmentation results for complex images.

Er. Amrinder Singh Brar et al. (2016) have proposed Fuzzy C means clustering for potato defect detection. Their proposed method was very effective in analyzing the defect.