جستجوی مقالات مرتبط با کلیدواژه « بخش بندی تصویر » در نشریات گروه « فنی و مهندسی »

In this research, a new approach for image segmentation based on genetic algorithm and random forest is presented. Image segmentation can be done using supervised learning. In this learning, there are a number of images from data set with their labels. In image segmentation, different parts of the image separate from each other. In this process, all the pixels in the image are given a label, so that the pixels with the same label have common characteristics with each other. To provide a model that can perform image segmentation, it is necessary to extract features from input images and perform segmentation using a suitable classifier and these features. Image feature extraction is done using image filters. In this research, a hybrid combination of 4 Gabor filter banks and Sobel, Prewitt, Canny edge, Scharr, Gaussian, median, and Roberts filters are used for effective feature extraction. One of the most important of these filters, which also has a degree of freedom, is the Gabor filter. This filter has a number of   hyperparameters that change the efficiency of the classifier by changing these hyperparameters. In this research, an attempt has been made to adjust these hyperparameters using genetic algorithm. The fitness function proposed in this research is f1-score. random forest classifier is utilized for image segmentation and classification. The results of the experiments show that the hyperparameters found by the genetic algorithm have been able to perform a satisfactory segmentation on data set.

Masonry structures comprise a large proportion of human-made building stocks around the world. In many cases, aged masonry structures have been found to be vulnerable to earthquakes and seismic loads. Due to the historical importance and vulnerable conditions of these structures, an efficient structural health monitoring system is required to detect every sign of degradation. Thus, a suitable restoration scheme could be taken into account. Manual visual inspection is one of the earliest monitoring schemes used to inspect these structures. Due to the limitations and dangers imposed by using human resources, new strategies are required to achieve this purpose. Recent developments in artificial Intelligence and computer vision have helped researchers develop a new generation of autonomous inspection systems. In the present study, we are going to use a deep learning model with an encoder-decoder architecture to automate crack detection in masonry structure images. In the current study, semantic segmentation is proposed as a detailed solution to accurately predict the location and condition of cracks in masonry images. In the development of the main model of the study, we used EfficientNet-B3 as the encoder while the decoder was defined according to U-Net’s expansion path in order to predict the accurate segmentation mask for the corresponding input images. For training and evaluation of the proposed model, a dataset composed of 115 images is generated and manually annotated. In the proposed method, transfer learning is used to train the model and the data augmentation techniques are implemented to achieve the optimal results on the present dataset. Furthermore, using the Dice-Coefficient loss function directly optimizes the model for F1-Score, which is the main evaluation parameter in semantic segmentation tasks. Finally, the evaluation demonstrated 81.444% Precision, 71.411% Recall, and 75.366% F1-Score for the never-seen test data. The study shows that the deep learning approach can be accurate and trustworthy for this task. Also, the limited number of training data and the complex background images in the dataset prove the robustness of the proposed model.

This paper presents a machine vision approach to detect the installation error of conrod bearing in four-cylinder engines. Since there are always human errors in engine production lines that would reduce the quality of the final product, it is vital to establish the intelligent machine vision systems in order to track the process of assembling parts and prevent installation errors. The proposed method is such that by taking an engine block image, it produces a binary image that the whole image is black and only the places where conrod bearings exist are white, and also it announces which bearings are present and which ones are not installed. To this end, firstly the images of 16 different bearing installation cases were captured by camera. Then, all images were analyzed with a combination of image processing methods including Gaussian filtering, image thresholding and segmentation, and morphological techniques including erosion and dilation. Finally, with applying if-then rules on the characteristics of the objects created in the image, its condition was decided. The results showed that the best threshold range for separating the bearing from other parts of image was between 110 and 245. Also a disk-shaped structuring element with the radius of 60 provided the best morphological tool to detect the bearings. In addition, the results demonstrated that the regions belonged to conrod bearings had an area between 30,000 and about 80,000 pixels. With using if-then rule, the different bearing installation cases in all images were successfully detected with 100% accuracy. The total time for analyzing each image was about 1 s. So the results showed that the proposed machine vision system provided a non-error and fast tool for detecting the existence of conrod bearings which can be served as a Poka-Yoke system in the engine production line.

Segmentation of facial color images is an essential step in the image processing and computer vision applications, such as face recognition, identity recognition, and analysis of facial plastic surgeries. One of the most important methods of facial image segmentation is clustering-based methods. The fuzzy c-means (FCM) clustering is an effective method in the image segmentation, but its sensitivity to initial values may cause that this algorithm fall and stuck into the local minima. To overcome this problem, the meta-heuristic algorithms, including Grey Wolf Optimization (GWO) and Whale Optimization Algorithm (WOA) have been used. Therefore, the main focus of this study is on the performance of the meta-heuristic algorithms in optimizing the FCM algorithm and their applications in the segmentation of facial color images. The objective function of the FCM algorithm is considered as a fitness function for meta-heuristic algorithms. This algorithm divides n vectors into C fuzzy groups and calculates the cluster center for each group. Also in this study, three color spaces (1) YCbCr, (2) YPbPr, and (3) YIQ have used as input data in optimization of the fitness function. After maximization of the membership function, segmentation of facial color images has been done on three database including, (1) Sahand University of Technology (SUT), (2) MR2, and (3) SCUTFBP. The result of segmentation show that convergence speed of the GWO and WOA methods is faster than other meta-heuristic algorithm, such as Genetic Algorithm (GA), Particle Swarm Optimization (PSO), Crow Search Algorithm (CSA), and Grasshopper Optimization Algorithm (GOA) and have a suitable performance in facial image segmentation.

This paper describes a method named hierarchical unsupervised segmentation using lossy data compression for synthetic aperture radar (SAR) images in which super pixels are used instead of pixels. In this paper, using a combination of several features such as edge, texture, and intensity, the super pixels are merged where the merging procedure is done by two steps. Since, many superpixel-based methods suffer from long run-times, the two step-proposed algorithm beside improving the accuracy, reduces the running-time. The first step is merging super pixels which there are not discriminative and clear edges between them. The merging of super pixels is done in the second step, if the length of data codes is minimized under a distinct distortion. The proposed algorithm is performed on SAR images and the simulation results demonstrate that the algorithm has a good accuracy and high processing speed.