The massive volume of images produced in recent years has made image retrieval one of the topics of research in the field of machine vision and image processing. The main challenge of content-based image retrieval systems is to extract the appropriate feature vector for image description to enable image retrieval effectively. In this research, a content-based image retrieval framework is introduced. The introduced feature vector is a combination of low-level features and mid-level features of the image. Extraction of low-level features of the image, including color, shape and texture, was performed using multi-level autocorrelation, discrete wavelet transform and fractal dimension analysis. Mid-level features are also extracted using the deep Boltzmann machine and by learning the low-level features of the image. The resulting feature vector is adjusted with 1K Corel database images and the performance of the proposed framework is also measured on 5K and 10K Corel databases. The best evaluation results are reported on 99.5%, 99.2% and 99.6% of the mentioned databases, respectively.

The massive volume of images produced in recent years has made image retrieval one of the topics of research in the field of machine vision and image processing. The main challenge of content-based image retrieval systems is to extract the appropriate feature vector for image description to enable image retrieval effectively. In this research, a content-based image retrieval framework is introduced. The introduced feature vector is a combination of low-level features and mid-level features of the image. Extraction of low-level features of the image, including color, shape and texture, was performed using multi-level autocorrelation, discrete wavelet transform and fractal dimension analysis. Mid-level features are also extracted using the deep Boltzmann machine and by learning the low-level features of the image. The resulting feature vector is adjusted with 1K Corel database images and the performance of the proposed framework is also measured on 5K and 10K Corel databases. The best evaluation results are reported on 99. 5%, 99. 2% and 99. 6% of the mentioned databases, respectively.

image retrieval is one of the most applicable image processing techniques, which have been used extensively. Feature extraction is one of the most important procedures used for interpretation and indexing images in content-based image retrieval (CBIR) systems. Reducing the dimension of feature vector is one of the challenges in CBIR systems. There are many proposed methods to overcome these challenges. However, the rate of image retrieval and speed of retrieval is still an interesting field of research. In this paper, we propose a new method based on the combination of Hadamard matrix, discrete wavelet transform (HDWT2) and discrete cosine transform (DCT) and we used principal component analysis (PCA) to reduce the dimension of feature vector and K-nearest neighbor (KNN) for similarity measurement.The precision at percent recall and ANR are considered as metrics to evaluate and compare different methods.Obtaining results show that the proposed method provides better performance in comparison with other methods.

The aim of this study is to characterize and find the location of geological boundaries in different wells across a reservoir. Automatic detection of the geological boundaries can facilitate the matching of the stratigraphic layers in a reservoir and finally can lead to a correct reservoir rock characterization. Nowadays, the well-to-well correlation with the aim of finding the geological layers in different wells is usually done manually. For a rather moderate-size field with a large number of wells (e. g., 150 wells), the construction of such a correlation by hand is a quite complex, labor-intensive, and time-consuming. In this research, the wavelet transform as well as the fractal analysis, with the aid of the pattern recognition techniques, are used to find the geological boundaries automatically. In this study, we manage to use the wavelet transforms approach to calculate the fractal dimension of different geological layers. In this process, two main features, the statistical characteristics as well as the fractal dimensions of a moving window, are calculated to find a specific geological boundary from a witness well through different observation wells. To validate the proposed technique, it is implemented in seven wells of one of the Iranian onshore fields in the south-west of Iran. The results show the capability of the introduced automatic method in detection of the geological boundaries in well-to-well correlations.

In machine learning, transferring and generalizing the knowledge learned from one domain to another is one of the important and basic capabilities. Since supervised learning is not complete, the use of other methods, such as self-supervised learning methods, can be very helpful in domain generalization. In this paper, we present a method that, in addition to classify original images in order to learn data labels in a supervised process, attempts to classify images resulting from the application of discrete wavelet transform on the original images by generating pseudo-labels for them. This extra work as a self-supervision task can lead to learn useful features and a general image representation for images of different domains, which can greatly help to improve the problem of domain generalization. In the following, by combining self-supervised methods such as jigsaw puzzles and guessing the rotation angle with discrete wavelet transform, we show that this combination can improve the results for the domain generalization problem. In this paper, we used the well-known PACS, VLCS and office-Home datasets to perform experiments, and the results show that our proposed method can work better than advanced and state-of-the-art domain generalization methods.

In many signal processing applications, such as EEG analysis, the non-stationary signal is often required to be segmented into small epochs. This is accomplished by drawing the boundaries of signal at time instances where its statistical characteristics, such as amplitude and/or frequency, change. In the proposed method, the original signal is initially decomposed into signals with different frequency bands using wavelet transform. The fractal dimension of the decomposed signal is calculated in a sliding window and the results are used as a feature for adaptive segmentation. A criterion is introduced in this paper to choose a proper length for the sliding window. Performance of the proposed method is compared with that of three other existing segmentation methods using synthetic and real EEG data. Simulation results show the high efficiency of the proposed method in signal segmentation.

In image inpainting, distorted and damaged parts of image or selected objects are removed or replaced with the appropriate information. In this article, image inpainting is performed by using frequency information of wavelet transform. The fill-in is done by diffusion of information of intact pixels into the damaged regions, which is begun from the outermost pixels and gradually the damaged region is reconstructed. To determine direction and the amount of diffusion, the geodesic path based image inpainting method is generalized by incorporating information of wavelet domain. The experimental results confirm superiority of the proposed method over the geodesic path based image inpainting method.

Analysis and interpretation of medical images are of clinical importance for medical diagnosis and treatment while they also have technical implications for computer vision and pattern recognition. In this context, one of the most fundamental issues is the detection of object boundaries, which is often useful for further processes such as organ/tissue recognition, image registration, motion analysis, measurement of anatomical and physiological parameters, etc. Although one of the best methods of edge detection is based on wavelet transform, the standard wavelet transform has its own shortcomings such as lack of shift invariant and lack of directional selectivity in sub-bands in multidimensional applications. The discrete complex wavelet transform, which is based on complex mother wavelet, not only overcomes these shortcomings but has acceptable redundancy and complexity as well. It is especially useful for multidimensional situations and for high accuracy applications such as medical image processing. In this paper, the shortcomings of ordinary wavelet transform are initially investigated and comparisons are made between the standard wavelet and the complex wavelet. Then, the discrete complex wavelet domain is applied for image enhancement and edge detection of noisy images. The simulation results show that our method exhibits a better performance, especially in noisy cases, as compared with the standard wavelet and spatial methods.