Pixels In Region Research Articles

A multidimensional chaotic image encryption algorithm based on the region of interest

Most image encryption algorithms encrypt the whole image, but only part of the data is important in the image. In this paper, we propose a multidimensional chaotic image encryption algorithm based on the region of interest (ROI). The histogram of oriented gradients (HOG) feature extraction and support vector machine (SVM) are used to separate the region of interest from the whole image. Then, the region of interest pixels is messed up by using the improved Henon sequence, Joseph sequence and the region of interest pixels are diffused by using the unified chaotic sequence to hide the sensitive information in the image, so as to achieve the purpose of private protection. Furthermore, the improved logistic sequence is used to hide the edge information of the target image to achieve the tradeoff between the secrecy of information and the complexity of encryption. A series of analyses are carried out including key space analysis, key sensitivity analysis, statistical analysis, information entropy analysis, analysis of the fixed-point obscuration analysis, quality analysis and image decryption for our encryption algorithm. Through experiments and comparisons, the proposed algorithm has good performance in encrypting image and coping with various invasions. The image encryption algorithm based on ROI has a good performance of security, moreover through the main encryption of ROI can effectively shorten the encryption time, so as to achieve the compromise of security and computational complexity.

Attributes guided facial image completion

Recently, facial image completion has witnessed substantial progress due to the increasing power of deep convolutional networks and generative adversarial nets. These methods try to reproduce the pixels in the missing regions of a facial image so that the completion results are indistinguishable from the original one. However, due to the multimodality of image contents, there are often many plausible ways to complete a facial image, which are equally good and meaningful. A user might be specifically interested in obtaining a particular completion result with certain desired attributes. To address this need, we propose a novel deep learning approach to facial image completion with multiple controllable attributes (e.g., gender and smiling). First, an attribute classification loss is proposed to guarantee the completed image should have some desired attributes. Second, our approach introduces a perceptual loss to model high-level semantic structures in a facial image, which greatly reduces structure distortions in the large missing regions. Third, an adversarial loss is employed to ensure the completion results are visually realistic. The three components cooperate well in our approach, forming an effective framework for high quality facial image completion with multiple controllable attributes. Experimental results on the CelebA datasets show our method not only achieves excellent performance on facial completion task, but also allows users to obtain specific completion results with facial attributes of interest.

Improvement of dehazing algorithm based on dark channel priori theory

The dehazing algorithm based on dark channel priori theory is an effective method for dehazing of single image. However, it still has shortcomings in some of special cases. In this work, a method is proposed to impair these disadvantages. By considering local smoothness of the sky region, pixels in the bright regions (such as light source, white object, etc.) of the haze image are first excluded from being mistaken as reference pixels, so that the estimation of subsequent parameters is more accurate. Meanwhile, the pixels of the most occurrences are selected to replace the ones in the neighborhood centered on the edge pixels, and the operation is carried out pixel by pixel along the edge contour in replace of the minimum filter. This makes the estimation of transmission near the edge in better agreement with the actual situation, and can effectively avoid the “white haze” which is an artifact caused by the minimum filtering in the areas with the depth of field changing dramatically. Compared with the previous methods, the proposed algorithm in this paper can restore images more clearly, with more image edge details retained, and effectively improve visual effect of the scene in haze weather.

Multilabel Sample Augmentation-Based Hyperspectral Image Classification

The quantity and quality of training samples have a great influence on the performance of most hyperspectral image classification approaches. However, in a real scenario, manually annotating a large number of accurate training samples is extremely labor-intensive and time-consuming. In this article, a multilabel training sample augmentation method is proposed. Instead of giving an exact label to each pixel, we just precisely label a small number of pixels by giving them a single label (called single-label samples) and annotate a large number of pixels in certain regions together by giving them multiple labels (called multilabel samples). Furthermore, in order to make full use of the multilabel training samples, a superpixel segmentation and recursive filtering-based method is proposed. The proposed method consists of the following major steps: recursive filtering-based feature extraction, superpixel-based segmentation, and spectral–spatial similarity-based mislabeled sample removal. Experimental results demonstrate that the proposed method can significantly improve the classification accuracy of multiple classifiers by using the multilabel training samples.

Color Transfer With Salient Features Mapping via Attention Maps Between Images

Color transfer is a useful image editing technology that transfers the color style of a reference image to a source image. In this paper, a novel automatic perception color transfer method based on salient features mapping was proposed. Firstly, the saliency maps of the reference image and source image are built to predict the human visual attention regions. Secondly, the gradient map is calculated by convolving the image pair with Sobel filter. Thirdly, the saliency map and gradient map are combined to produce a weighted attention map. Fourthly, by thresholding the attention map the reference image and the source image are segmented into regions of salient and non-salient pixels, and mean and standard deviation of these two regions are calculated in the three channels of the YUV space, respectively. Finally, on the salient and non-salient pixels an execution similar to Reinhard's is performed, and the two results are combined to generate the final output. The experiments result show the superiority of our method by its pleasant natural appearance with color similarity and image details preservation, high time efficiency and easy implementation.

Hyperspectral Image Target Detection via Weighted Joint K-Nearest Neighbor and Multitask Learning Sparse Representation

Multitask sparse representation method improves the detection performance by constructing multiple associated sub-sparse representation tasks and jointly learning multiple sub-sparse representation tasks, and this method can make use of the spectral information. However, the using of spatial information needs to be improved. This paper designs a hyperspectral image target detection method which can both make use of spectral and spatial information, that is a weighted joint k-nearest neighbor and multitask learning sparse representation method (WJNN-MTL-SR) is proposed. This method mainly consists of the following steps:1) using multitask sparse representation to obtain the representation residuals. 2) weighted joint k-nearest neighbor is used into the joint region of test pixels to obtain the weighted joint Euclidean distance. 3) a decision function, combining the weighted joint Euclidean distance and residuals of the multitask sparse representation, is used to get target detection result. Experimental results demonstrate that the proposed method show better detection performance than state-of-the-art methods.

A Spatially Constrained Probabilistic Model for Robust Image Segmentation.

In general, the hidden Markov random field (HMRF) represents the class label distribution of an image in probabilistic model based segmentation. The class label distributions provided by existing HMRF models consider either the number of neighboring pixels with similar class labels or the spatial distance of neighboring pixels with dissimilar class labels. Also, this spatial information is only considered for estimation of class labels of the image pixels, while its contribution in parameter estimation is completely ignored. This, in turn, deteriorates the parameter estimation, resulting in sub-optimal segmentation performance. Moreover, the existing models assign equal weightage to the spatial information for class label estimation of all pixels throughout the image, which, create significant misclassification for the pixels in boundary region of image classes. In this regard, the paper develops a new clique potential function and a new class label distribution, incorporating the information of image class parameters. Unlike existing HMRF model based segmentation techniques, the proposed framework introduces a new scaling parameter that adaptively measures the contribution of spatial information for class label estimation of image pixels. The importance of the proposed framework is depicted by modifying the HMRF based segmentation methods. The advantage of proposed class label distribution is also demonstrated irrespective of the underlying intensity distributions. The comparative performance of the proposed and existing class label distributions in HMRF model is demonstrated both qualitatively and quantitatively for brain MR image segmentation, HEp-2 cell delineation, natural image and object segmentation.

A Two-Stage Evolutionary Fuzzy Clustering Framework for Noisy Image Segmentation

This article presents a two-stage evolutionary fuzzy clustering framework for noisy image segmentation. It is a bi-stage system comprising a multi-objective optimization stage and a fuzzy clustering segmentation stage. In the multi-objective optimization stage, the fuzzy clustering on pixels in inhomogeneous regions is converted into a multi-objective problem, which can preserve image details while restraining noise. The multi-objective problem is decomposed into several sub-problems by the Tchebycheff approach. A trade-off can be obtained by optimizing these sub-problems simultaneously. In the fuzzy clustering segmentation stage, fuzzy clustering with the trade-off between preserving image details and restraining noise is performed on the whole observed image. To deal with this fuzzy clustering problem, an adaptive evolutionary fuzzy clustering algorithm with spatial information is proposed. Experiment results on synthetic and real images illustrate the effectiveness of the proposed framework for noisy image segmentation.

Image-derived index - as a paramether for the semi-quantitative assessment of global tubular renal function

Introduction. Radionuclide methods are used in the evaluation of different aspects of renal function. The aim of this study was to define and implement an index derived from scintigraphic image and to compare it with ortho-iodohippurate clearance values, an indicator of effective renal plasma flow. Material and Methods. A retrospective analysis included 67 adult patients, who underwent static renal scintigraphy and ortho-iodohippurate clearance. We computed the background-to-renal index that represents the ratio of average values of counts per pixel in the background regions and in the regions of kidneys. Patients were divided into three groups according to the values of measured effective renal plasma flow: group 1 - patients with mild renal dysfunction (reduction of effective renal plasma flow ? 20%, 17 patients), group 2 - moderate dysfunction (reduction of effective renal plasma flow 20-50%, 25), and group 3 - severe dysfunction (reduction of effective renal plasma flow 50-70%, 25 patients). Results. Subjects with mild effective renal plasma flow reduction had significantly lower background-to-renal index values compared to those with moderate and severe reduction [(0.105 ? 0.05) vs. (0.134 ? 0.056) vs. (0.275 ? 0.154), p < 0,001]. A significant linear correlation was found between background-to-renal index and effective renal plasma flow (r = 0.60, p < 0.0001). Sensitivity and specificity of the calculated cut-off value of 0.1335, for the differentiation between the patients with renal reduction higher and lower than 20%, were 62% and 100%, respectively (area under curve 0.85). Conclusion. The background-to-renal index is a simple method for semiquantitative estimation of global tubular renal function that can be a useful tool when determining patients for further in vitro assessment of global renal function.

Extraction of Gliomas from 3D MRI Images using Convolution Kernel Processing and Adaptive Thresholding

In this work, the evaluation of segmentation techniques towards the detection tumours or gliomas on BraTs 2018 data sets on T2 weighted 3D MRI images is carried out. Techniques experimented include pixel wise Linear Binary Pattern (LBP), region growing and Otsu thresholding to compare the efficiency of the same with proposed approach. In this work, 3D MRI images are initially subject to slicing from which about 10% from the top and bottom are normalized as those slices does not possess any significant details (empty slices). Further, the enhancement of tumour/gliomas regions is performed on normalized slices using high boost ‘convolution kernel processing in various trials with varying kernel sizes of high boost convolution kernel. The glioma regions enhanced using kernel are subject to local adaptive thresholding technique for extraction of gliomas from normalized slices. The detected gliomas are compared with the results of ground truth segmentation to determine the effectiveness of the proposed technique. It is observed in 80% of slices are successful in detection of gliomas both in case of Low Grade Glioma (LGG) and High Grade Glioma (HGG). By analyzing the results, it is observed that proposed method has outperformed with average Jaccard and Dice similarity recorded as 87.20% and 89.18% for HGG and where as in case of LGG it is found to be 83.77% and 87.74%.

Semantic segmentation of remote sensing image based on deep fusion networks and conditional random field

Image semantic segmentation refers to segmenting an image into several groups of pixel regions with different specific semantic meanings and identifying the categories of each region. In recent years, the common semantic segmentation methods that are based on Convolutional Neural Networks(CNN) have realised the pixel-to-pixel image semantic segmentation. They can avoid the problems of artificial design and selection of features in traditional image semantic segmentation methods. As a result of the pooling operation and lack of context information, the detailed information of images is neglected, the precision of the final image semantic segmentation result is low and the segmentation edge is inaccurate. Therefore, this study proposes a semantic segmentation method for remote sensing image on the basis of Deep Fusion Networks(DFN) combined with a conditional random field model.The method initially builds a DFN model in a Fully Convolutional Network(FCN) framework with a deconvolutional fusion structure.On the one hand, the multiscale features can be extracted through the deep networks, which can avoid the artificial design and selection of features to improve the generalisation ability of the model. On the other hand, the multiscale information is used in the model with the help of the deconvolutional fusion structure. The processing accuracy of the model is also improved by fusing the shallow detail information and deep semantic information. Fundamentally, the fully connected conditional random field is introduced to supplement the spatial context information towards precisely locating the boundary and obtaining final semantic segmentation results.From this study, we can draw the following conclusions:(1)With the increase in the depth of the fusion layer, detailed information becomes abundant, the semantic segmentation results become refined and the edge contour becomes close to the label image;(2) The fully connected conditional random field model synthesises the global and local information of the remote sensing image and further improves the efficiency and accuracy of the final semantic segmentation results.

Toward an Optimized Neutrosophic k-Means With Genetic Algorithm for Automatic Vehicle License Plate Recognition (ONKM-AVLPR)

The present paper proposes a new methodology for license plate (LP) recognition in the state of the art of image processing algorithms and an optimized neutrosophic set (NS) based on genetic algorithm (GA). First of all, we have performed some image processing techniques such as edge detection and morphological operations in order to utilize the (LP) localization. In addition, we have extracted the most salient features by implementing a new methodology using (GA) for optimizing the (NS) operations. The use of (NS) decreases the indeterminacy on the (LP) images. Moreover, k-means clustering algorithm has been applied to segment the (LP) characters. Finally, we have applied connected components labeling analysis (CCLA) algorithm for identifying the connected pixel regions and grouping the appropriate pixels into components to extract each character effectively. Several performance indices have been calculated in order to measure the system efficiency such as accuracy, sensitivity, specificity, dice, and jaccard coefficients. Moreover, we have created a database for all detected and recognized (LP) for testing purposes. Experimental results show that the proposed methodology has the ability to be suitable for both (Arabic –Egyptian) and English (LP). The proposed system achieves high degree of recognition accuracy for the whole system according to the following case studies; (i) for a high resolution Egyptian (LP), the proposed system achieves about 96.67% accuracy of correct recognition, (ii) for a low resolution-corrupted English (LP), the proposed system achieves about 94.27% accuracy. In addition, we have applied the proposed system on some sort of image disturbance i.e. (flash in image, external noise, and illumination variation), the proposed system achieves about 92.5% accuracy of correct identification. However, traditional methods achieve about 79% accuracy of correct identification in the presence of such image degradations. This reflects how the proposed system is generalized, optimized, and proposes high degree of recognition accuracy.

Evaluation of tree-base data mining algorithms in land used/land cover mapping in a semi-arid environment through Landsat 8 OLI image; Shiraz, Iran

Land Use Land Cover (LULC) mapping has been used in different environmental applications including disaster management, risk analysis, heat island mapping, and the effects of urbanization on environmental changes such as floods and droughts in the recent decade. The earth’s natural surface coverage including urban infrastructure, surface vegetation, bare soil can be identified with LULC maps. Besides, LULC is one of the most important tasks in natural hazards, planning activities, resource management, and global monitoring studies. The present study aimed to improve the level of land used and land cover change mapping in a semi-arid environment using Landsat 8 Operational Land Imager (OLI) image. In this research, several tree-based algorithms for LULC mapping are used and compared. These algorithms are used in a combination model implementation of WEKA 3.8 and R programming language to provide a method of a fit-for-purpose algorithms for LULC mapping. It is found that Reduced Error Pruning Tree (REP Tree) is the best approach according to the training and test datasets which are four classes including the build-up, soil, roads, and vegetation region pixels in a semi-arid environment. For the training dataset, the best values of Overall Accuracy (OA), Mean Absolute Error (MAE), and Root Mean Square Error (RMSE) is determined, 99.7313, 0.002 and 0.0354 as well as for the test dataset are 99.7313, 0.002 and 0.0354, respectively. Finally, trained models of the implemented data mining algorithms are used for the classification of a different dataset without using any training data where the Logical Analysis of Data Tree (LAD Tree) algorithm shows the best performance in terms of OA, MAE, and RMSE with values of 99.2815, 0.0057, and 0.0557 respectively.

Block-segment search of local extrema of images based on analysis of brightnesses of related pixels and areas

The aim of the work is to develop an algorithm for extracting local extremes of images with low computational complexity and high accuracy. The known algorithms for block search for local extrema have low computational complexity, but only strict maxima and minima are distinguished without errors. The morphological search gives accurate results, highlighting the extreme areas formed by non-severe extremes, however, it has high computational complexity. The paper proposes a block-segment search algorithm for local extremums of images based on an analysis of the brightness of adjacent pixels and regions. The essence of the algorithm is to search for single-pixel local extremes and regions of uniform brightness, comparing the values of their boundary pixels with the values of the corresponding pixels of adjacent regions: the region is a local maximum (minimum) if the values of all its boundary pixels are larger (smaller) or equal to the values of all adjacent pixels. The developed algorithm, as well as the morphological search algorithm, allows detecting all single-pixel local extremes, as well as extreme areas, which exceeds the block search algorithms. At the same time, the developed algorithm in comparison with the morphological search algorithm requires much less time and RAM.

Detection of abnormalities in mammograms using deep features

The mortality rate of breast cancer can be reduced by early diagnosis and treatment. The computer-aided diagnosis (CAD) systems can effectively help physicians identify the early stage of breast cancer. The primary tool for such systems is mammograms. Usually, these images lack high quality. Furthermore, due to the irregular shape of masses, their size variability, and the apparent similarity of the masses and other dense regions of the breast tissue, it is difficult to detect and diagnose the masses. Although many image processing techniques have been presented for diagnosis of breast masses, they have not been quite successful, and this problem has been retained as a challenge yet. In this paper, a method for classifying breast tissues into normal and abnormal (i.e., cancerous) is proposed, which is based on a deep learning approach. It mainly contains a new convolutional neural network (CNN) and a decision mechanism. After a preprocessing phase, a block around each pixel is fed into a trained CNN to determine whether the pixel belongs to normal or abnormal tissues. This results in a binary map for the input suspicious tissue. Afterward, as a decision mechanism, a thresholding technique is applied to a central block on the produced binary map to label it. The new architecture of the CNN, the training scheme of the network, employing the CNN for classifying the pixels of the suspicious regions rather than the entire input, exerting an effective decision mechanism on the output of the CNN and the learning of the model parameters, helped us achieve superior results compared to state-of-the-art methods by reaching 95 and 94.68 percent for AUC and accuracy, respectively.

Regional air trapping in acute exacerbation of obstructive lung diseases measured with electrical impedance tomography: a feasibility study.

Since bronchial abnormalities often exhibit spatial non-uniformity which may be not correctly assessed by conventional global lung function measures, regional information may help to characterize the disease progress. We hypothesized that regional air trapping during mechanical ventilation could be characterized by regional end-expiratory flow (EEF) derived from electrical impedance tomography (EIT). Twenty-five patients suffering from chronic obstructive pulmonary disease (COPD grade 3 or 4) or severe asthma with acute exacerbation were examined prospectively. Patients were ventilated under assist-control mode. EIT measurements were conducted before and one hour after inhaled combined corticosteroid and long-acting β2 agonist, on two consecutive days. Regional EEF was calculated as derivative of relative impedance for every image pixel in the lung regions. The results were normalized to global flow values measured by the ventilator. Regional and global EEF were highly correlated (P<0.00001) and regional effects of medication and disease progression were visible in the regional EEF maps. The sums of regional EEF in lung regions were 3.8 [2.0, 5.1] and 3.6 [1.9, 4.5] L/min in COPD patients before and after medication (median [lower, upper quartiles]; P=0.37). The corresponding values in asthma patients were 3.0 [2.5, 4.2] and 2.2 [1.7, 3.2] L/min (P<0.05). Histograms of regional EEF showed high spatial heterogeneity of EEF before medication. After one day of treatment, the histograms exhibited less heterogeneous and a decrease in EEF level. Regional EEF characterizes air trapping and intrinsic PEEP, which could provide diagnostic information for monitoring the disease progress during treatment.

Quantitative Detection of Contact Force Chains in a Model Particle Assembly Using Digital RGB Photoelastic Measurements

This paper explains a newly developed technique for identifying particles belonging to the contact force chains in a model particle assembly. This was accomplished through digital image analysis of the color images obtained from photoelastic measurements. Descriptive statistics for RGB color intensities of pixels in a central square region inside the particle were digitally measured to determine the particle force, as opposed to the traditional method of counting fringe orders. The relationships between the descriptive statistics of the RGB color intensities and the applied forces were analyzed. The developed image analysis technique was applied to digital photoelastic images of an assembly of stacked particles. This technique could embed the numeric data of the particle forces in the digital photoelastic images which provide comprehensive information for a quantitative analysis of the progressive evolution of the soil arching phenomena in a particulate media.

Fusing texture, edge and line features for smoke recognition

To improve recognition accuracy, the authors fuse texture, edge and line information to propose a feature extraction method for smoke recognition. The Canny operator is proposed to generate an edge image from an original image, and then adopt the Hough transform to extract straight lines from the edge image. The lines are rasterised to generate a discrete line image and two local patterns are proposed for the edge and line images. The first one is local boundary summation pattern (LBSP) that computes the sum of binary pixel values along the boundary of a local region around a centre pixel. The second one is called local region summation pattern (LRSP) that sums up the binary values of pixels in a local region around the centre pixel. Besides LBSP and LRSP, LBPs with three mapping modes (LBP_M3) to achieve traditional texture information are also extracted. Finally, the authors concatenate the histograms of LBP_M3, LBSP and LRSP to generate a feature vector, and use support vector machine for classifying and testing. Experiments show that authors’ method outperforms most of existing traditional methods for smoke recognition. Although this method has low dimensional features, it also obtains good performance for multi-class texture classification.

Target Image Mask Correction Based on Skeleton Divergence

Traditional approaches to modeling and processing discrete pixels are mainly based on image features or model optimization. These methods often result in excessive shrinkage or expansion of the restored pixel region, inhibiting accurate recovery of the target pixel region shape. This paper proposes a simultaneous source and mask-images optimization model based on skeleton divergence that overcomes these problems. In the proposed model, first, the edge of the entire discrete pixel region is extracted through bilateral filtering. Then, edge information and Delaunay triangulation are used to optimize the entire discrete pixel region. The skeleton is optimized with the skeleton as the local optimization center and the source and mask images are simultaneously optimized through edge guidance. The technique for order of preference by similarity to ideal solution (TOPSIS) and point-cloud regularization verification are subsequently employed to provide the optimal merging strategy and reduce cumulative error. In the regularization verification stage, the model is iteratively simplified via incremental and hierarchical clustering, so that point-cloud sampling is concentrated in the high-curvature region. The results of experiments conducted using the moving-target region in the RGB-depth (RGB-D) data (Technical University of Munich, Germany) indicate that the proposed algorithm is more accurate and suitable for image processing than existing high-performance algorithms.

Phase unwrapping for Doppler spectral domain optical coherence tomography flow measurement.

Doppler optical coherence tomography (OCT) offers additional flow velocity information, which extends the application of OCT. Phase wrapping is the inherent problem that limits measureable range of Doppler OCT. We propose a phase unwrapping method which is suitable for correcting phase in Doppler OCT images. Points (pixels) in flow region are divided into groups according to the radial distance. Points in the same group are supposed to have close velocity. Phase unwrapping algorithm begins at the boundary layer group and is performed sequentially toward the center. Using the proposed criterion, points in a group are separated into two categories, signal points and noise points. Wrapping rounds are determined for signal points phase unwrapping. Mean value of the corrected signal points replaces the noise points for noise reduction. The method is validated with capillary tube flow phantom and in vivo blood flow.