Figure - available from: Multimedia Tools and Applications
This content is subject to copyright. Terms and conditions apply.
Source publication
Surface defect inspection plays a vital role in leather manufacturing. Current practice involves an expert to inspect each piece of leather individually and detect defects manually. However, such a manual inspection is highly subjective and varies quite considerably from one assorter to another. Computer vision system for natural material like leat...
Similar publications
The non-contact continuous monitoring of biomarkers comprising breathing detection and heart rate are essential vital signs to evaluate the general physical health of a patient. As compared to existing methods that need dedicated equipment (such as wearable sensors), the radio frequency (RF) signals can be syn-thesised to continuously monitor breat...
Citations
... Most traditional machine vision methods for defect detection require engineers to manually design the feature-extraction algorithms, which rely on trial and error and the performance varies case by case [35]. To minimize reliance on human experience, machine learning techniques such as SVM [36,37], KNN [38], and random forest [39] have been widely adopted to enhance the robustness of defect identification. However, machine learning algorithms are still considered less suitable for complex industrial applications. ...
Accurate detection of injection molding defects plays a crucial role in effective quality control of injection-molded products. This paper presents a U-Net training strategy, namely Semi-Supervised Auxiliary Learning (SSAL), that incorporates the morphological information of the inspected products into the U-Net network. Through the utilization of defect-free samples in semi-supervised auxiliary learning and the development of a well-designed training strategy, the U-Net network’s ability to localize injection molding defects is significantly enhanced. The introduction of defect-free samples to expand the dataset and the utilization of pre-trained models effectively reduces the risk of overfitting and the dependency on labeled datasets. Additionally, we have created an open-source dataset of injection-molded products with flash defect. The proposed SSAL has been evaluated against PSPNet, DeepLabv3+, and several U-Net methods and demonstrated superior performance. The mIoU (mean intersection over union), Recall, and mPrecision (mean precision) scores reach the values of 89.40%, 94.95%, and 93.32% respectively. This study precisely segments pixels depicting flash defect with complex contours, providing decision support for subsequent adjustment of injection molding machine production parameters and quality control in injection-molded product production.
... Most current practices in industry involve experts inspecting individually each piece produced and detecting defects manually (Jawahar et al., 2021). Particularly in the aerospace manufacturing industry, visual inspection still dominates the testing of parts including engine blades, accounting for approximately 90% of all inspections (Aust et al., 2021). ...
The need to improve reliability and support decision-making in manufacturing has drawn attention to the application of diagnostic and decision-support tools. Particularly in the investment casting industry, data-driven methods can be the enabler for process diagnostics and decision support. Images from the microscopic examination in the investment casting process are used as data input, to detect defects in produced pieces. The microscopic examination usually relies solely upon the ability of the operator to determine whether an image from the microscope contains a defect. Therefore, an effective strategy for this decision-making process is crucial to improve the reliability of the examination. The use of the machine learning classifier Random Forest is introduced to derive predictions on the existence of a defect in the input image. This work focuses on employing machine learning tools for image recognition and the developed approach constitutes a decision support model to assist the operator and improve the reliability of their assessment.
... Many types of texture feature based methods already exist and the most widely used is based on Haralick's texture features. In this method features such as correlation, energy, entropy, inertia, sum average, inverse difference moment, sum variance, difference average, sum entropy, difference entropy, difference variance, information measure of correlation 1 and of correlation 2 were extracted from gray level cooccurrence matrix (GLCM) and used as a texture feature vector [28,37]. Rotation-invariant image feature descriptor called Angular Radial Partitioning (ARP) was proposed for natural image retrieval applications [15,16]. ...
... Moganam et al. [46] utilized an intelligent system with machine learning models with GLCM features and achieved an accuracy of 99% for 1232 images including various major defects such as folding marks, grain off, growth marks, and pin holes. Malathy et al. [37] performed an improved fast convergent PSO model to extract the GLCM features and attained 88.64% of accuracy for random forest model. Liong et al. [52] improved the classification process for defects like tick bite using decision trees, Support Vector Machines, nearest neighbour achieving and accuracy of 84%. ...
Leather quality inspection is essential in determining the usable area of the material. As leather is a natural substance, surface defects can have a significant impact on its overall quality and reduce its usability. The automatic identification of surface defects in leather holds great importance in the inspection process. This study presents an innovative method called the Fourier Angular Radial Partitioning (FARP) algorithm for extracting features, specifically tailored for the identification of surface defects in leather. A cutting-edge industrial prototype machine vision system is designed with innovative capabilities to acquire high-quality entire leather surface image accurately. The FARP algorithm leverages a combination of spatial and radial distributed invariant feature descriptors obtained from the magnitude of the Fourier Transform. Furthermore, by partitioning the image into multiple sub-regions enables the FARP to extract features to effectively analyze both prominent flaws like cuts, scars and subtle imperfections like pinholes. The performance of the proposed FARP algorithm is compared to Gray Level Co-occurrence method and Spatial domain features. Correlation analysis is conducted on the extracted features from these three methods to identify the optimal feature set. Leather defects are classified using a multinomial logistic regression model and an ensemble classifier approach with random forest. Various measures, including accuracy, specificity, sensitivity, F-score, Mathew Correlation Coefficient, and ROC analysis using Z-test, are employed for a comprehensive evaluation. The experimental results indicate that the random forest and the proposed FARP feature set, achieves a remarkable classification accuracy of 88.67% and a notable area under the ROC curve of 0.875. This intelligent solution, which integrates FARP with the Random Forest classifier, surpasses the performance of manual expert leather defect classification, highlighting its superior effectiveness.
... The fitness function plays a key role in defect detection, requiring the ability to differentiate between defective and nondefective areas and identify various types of defects [24]. For instance, in a study on leather defect detection, a modified fitness function using selective-band Shannon entropy improved segmentation efficiency [25]. The velocity setting also influences algorithm convergence and search efficiency [26]. ...
This paper presents an advanced methodology for defect prediction in radiographic images, predicated on a refined particle swarm optimization (PSO) algorithm with an emphasis on fluctuation sensitivity. Conventional PSO models with stable velocity are often beleaguered with challenges in precisely pinpointing defect regions in radiographic images, attributable to the lack of a defect-centric approach and the propensity for premature convergence. The proposed fluctuation-sensitive particle swarm optimization (FS-PSO) model, distinguished by an approximate 40% increase in particle entrapment within defect areas and an expedited convergence rate, necessitates a maximal additional time consumption of only 2.28%. The model, also characterized by reduced chaotic swarm movement, enhances efficiency through the modulation of movement intensity concomitant with the escalation in swarm size. The FS-PSO algorithm’s performance was rigorously evaluated via a series of simulations and practical blade experiments. The empirical findings evince that the FS-PSO model substantially outperforms the conventional stable velocity model, particularly in terms of shape retention in defect extraction.
... Other works use simple NN as multilayer perceptrons (MLP) [39,40]. For the classification task, it turned out that Support Vector Machine (SVM) is the most used algorithm; it was used in 10 works [19,26,30,33,39,[41][42][43][44][45][46]. Apart from SVM, other machine learning (ML) algorithms deserve to be mentioned, e.g., k-Nearest-Neighbors (k-NN) presented in five papers [26,33,45,47,48], clustering algorithms [11,49], and random forest [45]. ...
Quality inspection in the industrial production field is experiencing a strong technological development that benefits from the combination of vision-based techniques with artificial intelligence algorithms. This paper initially addresses the problem of defect identification for circularly symmetric mechanical components, characterized by the presence of periodic elements. In the specific case of knurled washers, we compare the performances of a standard algorithm for the analysis of grey-scale image with a Deep Learning (DL) approach. The standard algorithm is based on the extraction of pseudo-signals derived from the conversion of the grey scale image of concentric annuli. In the DL approach, the component inspection is shifted from the entire sample to specific areas repeated along the object profile where the defect may occur. The standard algorithm provides better results in terms of accuracy and computational time with respect to the DL approach. Nevertheless, DL reaches accuracy higher than 99% when performance is evaluated targeting the identification of damaged teeth. The possibility of extending the methods and the results to other circularly symmetrical components is analyzed and discussed.
... Figure 5 shows a peek into one image from each class before pre-processing To highlight the features and also to bring uniformity in contrast and brightness, various image processing techniques like cropping, resizing, applying masks, image smoothing and blending were applied. Wiener filter provides an optimal tradeoff in restoring the effects of unequal brightness and contrast as well as noise smoothing in the DR images [27]. a look at the same images in each class, it is seen that the features are visible clearly. ...
... To highlight the features and also to bring uniformity in contrast and brightness, various image processing techniques like cropping, resizing, applying masks, image smoothing and blending were applied. Wiener filter provides an optimal tradeoff in restoring the effects of unequal brightness and contrast as well as noise smoothing in the DR images [27]. a look at the same images in each class, it is seen that the features are visible clearly. ...
Diabetic Retinopathy (DR) is an eye condition that mainly affects individuals who have diabetes and is one of the important causes of blindness in adults. As the infection progresses, it may lead to permanent loss of vision. Diagnosing diabetic retinopathy manually with the help of an ophthalmologist has been a tedious and a very laborious procedure. This paper not only focuses on diabetic retinopathy detection but also on the analysis of different DR stages, which is performed with the help of Deep Learning (DL) and transfer learning algorithms. CNN, hybrid CNN with ResNet, hybrid CNN with DenseNet are used on a huge dataset with around 3662 train images to automatically detect which stage DR has progressed. Five DR stages, which are 0 (No DR), 1 (Mild DR), 2 (Moderate), 3 (Severe) and 4 (Proliferative DR) are processed in the proposed work. The patient’s eye images are fed as input to the model. The proposed deep learning architectures like CNN, hybrid CNN with ResNet, hybrid CNN with DenseNet 2.1 are used to extract the features of the eye for effective classification. The models achieved an accuracy of 96.22%, 93.18% and 75.61% respectively. The paper concludes with a comparative study of the CNN, hybrid CNN with ResNet, hybrid CNN with DenseNet architectures that highlights hybrid CNN with DenseNet as the perfect deep learning classification model for automated DR detection.
... As an alternative to texture analysis, histogram thresholding, clustering, and so on, various biologically inspired algorithms were explored in image segmentation. Jamadar et al. [52] developed a fast convergence Particle Swarm Optimization algorithm (FCPSO) for segmenting defective regions in complex leather images. The Particle Swarm Optimization (PSO) is a heuristic algorithm loosely inspired by birds flocking in search of food. ...
... Each GLCM has an associated angle and displacement, related to the direction and frequency that will be represented by this GLCM. The most successful and highly used handcrafted texture features in the literature are Haralick features [52] derived from GLCM. Based on GLCM, Haralick calculated 14 statistics features [51]: energy, entropy, contrast, uniformity, correlation, variance, sum average, sum variance, sum entropy, difference variance, difference average, difference entropy, correlation information measure, and maximum correlation coefficient. ...
... Notably, these datasets contain at most 10 categories of defects, but most of them include three to four categories. Although the dataset used by Jawahar et al. [52,61,62] contains 10 categories of defects, it is divided into two types: defect and no defects. All datasets used in the literature [14,41,63,66,68] contain only one defect, which is essentially a binary classification. ...
Machine-vision-based surface defect inspection is one of the key technologies to realize intelligent manufacturing. This paper provides a systematic review on leather surface defect inspections based on machine vision. Leather products are regarded as the most traded products all over the world. Automatic detection, location, and recognition of leather surface defects are very important for the intelligent manufacturing of leather products, and are challenging but noteworthy tasks. This work investigates a large amount of literature related to leather surface defect inspection. In addition, we also investigate and evaluate the performance of some edge detectors and threshold detectors for leather defect detection, and the identification accuracy of the classical machine learning method SVM for leather surface defect identification. A detailed and methodical review of leather surface defect inspection with image analysis and machine learning is presented. Main challenges and future development trends are discussed for leather surface defect inspection, which can be used as a source of guidelines for designing and developing new solutions in this field.
... In intelligent systems, many scheduling and decision problems can be transformed into optimization problems, such as energy system [33], image processing [43], fault detection [18] and path planning [34]. With the development of technology, the optimization problems in intelligent systems are fluctuant, dynamic, nonlinear and incomplete. ...
Although particle swarm optimization (PSO) algorithm shows excellent performance in solving optimization problems, how to balance exploration and exploitation is still a crucial problem. In this paper, an adaptive comprehensive learning particle swarm optimization with spatial weighting (APSO-SW) is proposed. In APSO-SW, in order to increase the population diversity, the Euclidean distance between each particle and the global optimum is calculated in each generation and the particles in the whole population select their exemplars learning weight according to their Euclidean distance adaptively. Therefore, not only different particles have various learning weight, but also the same particle can select learning weight adaptively in different generations. In addition, for the purpose of improving the convergence property, the terminal elimination strategy is used. In terminal elimination strategy, the population can delete inferior particles and add preferable particles dynamically during the process of evolution. The comparisons among APSO-SW and other 7 state-of-the-art PSO variants on the CEC2013 and CEC2017 test suites reflect that APSO-SW is a trustable and remarkable optimization algorithm for solving various types problems. Furthermore, extensive experiments confirm the validity of our method.
... Structural analysis has always dealt with some rules, unique primitives [23] and these schemes always suffer complications of identifying unique patterns. So defect identification using structural approaches is suitable for the ones used in printed circuit board inspection [21] and are not suitable for real textured images like leather. Therefore, generally, all the defect detection process performs few statistical calculations to declare defects in the statistical process. ...
... In order to correct the irregular illuminated leather hides the lower Sensor Exview HAD CCD [52] Chris C. Bowman, ( 1996) [6 ] Jang-Woo Kwon (2004) [8] Malathy ( 2014) [9 ] Fig. 4 Image Acqusition Schemes used by various Researchers. Chris C. Bowman, (1996) [19], Jang-Woo Kwon (2004) [31], Malathy (2014) [21]. ...
... Bong et al. [5] proposed an SVM based vision based defect detection in leather images. Malathy et al. [20,21] presented an optimized defect segmentation using particle swarm optimization and classified the leather defects using neural network. ...
Increasing consumer quality awareness and increase in consumer wealth drives the market demand for high quality leather and leather products. Reliable and effective detection and classification of leather surface defects is of profound significance to tanneries and industries where leather is a major raw material for leather accessories and leather parts manufacturers. This paper presents a methodical and a detailed review of the leather surface defects detection methods starting from leather image acquisition, leather image processing, feature extraction and classification for defect detection. Firstly, we introduce the fundamentals of leather image acquisition and various related image processing methods, feature extraction and classification for the defect inspection. Next, the existing datasets and summary of the recent methodologies used in this field are discussed. Finally, the challenges and suggested improvements to further the development of the application of advanced Machine Learning and Deep Learning in this field are discussed. Deep learning algorithms are shown to have a great potential for leather surface defect detection and can help prepare a robust system that would greatly guarantee quality leather and provide monetary wealth from such leather products. Finally, research guidelines are presented to fellow researchers regarding data augmentation, leather defect detection models which need to be investigated in the future to make progress in this crucial area of research.
... Viana et al. presented an empirical evaluation of support vector machine against AdaBoost and MLP, for solving the leather defect classification problem [15]. Supervised classification using the multi-layer perceptron (MLP), Decision trees (DT), SVM, Naïve Bayes, KNN, and Random forest (RF) classifiers were used to classify the defective and non-defective leather regions [16]. The neural network classifier is proposed by multilayer perceptron neural networks for recognizing leather defects like open cut, closed cut, and Fly Bite [17]. ...
... Here the T corresponds to the "target" labels and the O stands for computed probability from the SoftMax function. The cross-entropy based cost function is (16) minimized for the given training data set using the stochastic gradient descent method by iteratively updating the weight matrix until the specified number of epochs or desired cost threshold is reached. ...
Modern leather industries are focused on producing high quality leather products for sustaining the market competitiveness. However, various leather defects are introduced during various stages of manufacturing process such as material handling, tanning and dyeing. Manual inspection of leather surfaces is subjective and inconsistent in nature; hence machine vision systems have been widely adopted for the automated inspection of leather defects. It is necessary develop suitable image processing algorithms for localize leather defects such as folding marks, growth marks, grain off, loose grain, and pinhole due to the ambiguous texture pattern and tiny nature in the localized regions of the leather. This paper presents deep learning neural network-based approach for automatic localization and classification of leather defects using a machine vision system. In this work, popular convolutional neural networks are trained using leather images of different leather defects and a class activation mapping technique is followed to locate the region of interest for the class of leather defect. Convolution neural networks such as Google net, Squeeze-net, RestNet are found to provide better accuracy of classification as compared with the state-of-the-art neural network architectures and the results are presented.
Graphical Abstract
