Vectorization Method of Satellite Images Based on Their Decomposition by Topological Features

The authors propose a method for automatic classification of spatial objects in images under conditions of a limited data set. The stability of the method to distortions appearing in images due to natural phenomena and partial overlap of urban infrastructure objects is investigated. High classification accuracy, when using existing approaches, requires a large training sample, including data sets with distortions, which significantly increases computational complexity. The paper proposes a method for a two-step topological analysis of images. Topological features are initially extracted by analyzing the image in the brightness range from 0 to 255, and then from 255 to 0. These features complement each other and reflect the topological structure of the object. Under certain deformations and distortions, the object preserves its structure in the form of extracted features. The advantage of the method is a small number of patterns, which reduces the computational complexity of training compared to neural networks. The proposed method is investigated and compared with the modern neural network approach. The study was performed on a DOTA dataset (Dataset for Object deTection in Aerial images) containing images of spatial objects of several classes. In the absence of distortion in the image, the neural network approach showed a classification accuracy of over 98 %, while the proposed method achieved about 82 %. Further distortions such as 90 degree rotation, 50 % narrowing and 50 % edge truncation and their combinations were applied in the experiment. The proposed method showed its robustness and outperformed the neural network approach. In the most difficult combination of the test, the decrease in classification accuracy of the neural network was 46 %, while the described method showed 12 %. The proposed method can be applied in cases with a high probability of distortion in the images. Such distortions arise in the field of geoinformatics when analyzing objects of various scales, under different weather conditions, partial overlap of one object with another, in the presence of shadows, etc. It is possible to use the proposed method in vision systems of industrial enterprises for automatic classification of the parts that belong to superimposed objects.

This paper presents an approach for the classification of satellite images by recognizing various objects in them. Satellite images are rich in geographical information that can be used in a number of useful ways. The proposed system classifies satellites images by extracting different objects from the images. Our object recognition mechanism extracts attributes from satellite images under two domains namely: color pixels’ organization and pixel intensity. The extracted attributes aid in the identification of objects lying inside the satellite images. Once we are able to identify objects, we proceeded further to classify satellite images with the help of decision trees. The system has been tested for a number satellite images acquired from around the globe. The objects in the images have been further subdivided into different sub categories to improve the classification and prediction process. This is a novel approach which is not using any image processing techniques but is utilizing the extracted features to identify objects and then using these objects to classify the satellite images.

Hyper spectral image are used in various applications such as geological systems, geo sciences and astronomy. These images are acquired using remote sensing. Remote sensing is the process of getting information about an object without making any physical contact with the object. Satellite Images referred as hyper spectral images are the most used images in remote sensing and are of more interest to find out the classification of objects in those images. The classification can give us the important factors like vegetation, buildings, roads and more. Satellite images can be of assistance in supervision of effects due to natural disasters, to recognize mining areas which are hidden from human view, biodiversity examination, rural and urban environment detection for analysis, etc. However, occasionally the Satellite images acquired can be affected by unforeseen distortions, artificial unwanted structures called artifacts that are formed by the tool itself or sometimes due to the diverse pre-processing procedures involved. Optimization algorithms in combination with Image processing methods are used to classify the objects in satellite images for easy perception and analysis. In this paper, various optimization techniques like particle swarm optimization (PSO), DPSO, HSO, and Proposed MFA optimization algorithms are compared to obtain optimal classification of objects in a satellite image.

Background subtraction is a useful and effective method for detecting moving objects in video images. Since this method assumes that image variations are caused only by moving objects (i.e., the background scene is assumed to be stationary), however, its applicability is limited. In this paper, we propose a background subtraction method that robustly handles various changes in the background. The method learns the chronological changes in the observed scene's background in terms of distributions of image vectors. The method operates the subtraction by evaluating the Mahalanobis distances between the averages of such image vectors and newly observed image vectors. The method we propose herein expresses actual changes in the background using a multi-dimensional image vector space. This enables the method to detect objects with the correct sensitivity. We also introduce an eigenspace to reduce the computational cost. We describe herein how approximate Mahalanobis distances are obtained in this eigenspace. In our experiments, we confirmed the proposed method's effectiveness for real world scenes.

Remote sensing imagery is a very interesting topic for researchers, especially in the fields of image and pattern recognition. Remote sensing images differ from ordinary images taken with conventional cameras. Remote sensing images are captured from satellite photos taken far above the Earth's surface. As a result, objects in satellite images appear small and have low resolution when enlarged. This condition makes it difficult to detect and recognize objects in remote-sensing images. However, detecting and recognizing objects in these images is crucial for various aspects of human life. This paper aims to address the problem of remote sensing image quality. The method used is a convolutional neural network. The results show the proposed method can improve PSNR and SSIM compared to previous methods

The problem of interpretation of spatial data on satellite images is considered in the article. It is proposed to use the decomposition of images by topological features to highlight objects of interest, global and detailed structures on satellite images. The description of the method and the features of its implementation for creating a software product with effective algorithms for processing big data are given. The functionality of the developed software, which includes the classification of objects on satellite images, segmentation, binarization, noise removal is described. It is shown that these algorithms are built on a single theoretical basis in the form of a topological decomposition. Examples of using the program for segmentation and binarization of satellite images from urban neighborhoods are demonstrated.

In similarity retrievals of trademark images, evaluation of similarity for essential objects which show products or services is required. In order to examine similarity of local objects in images, it is necessary to extract the objects; however, it is difficult to extract essential objects correctly from raster-based images. On the other hand, since vector graphics independently describe information to every object in an image, vector-based images could be effective to evaluate the similarity. To enhance performance of content-based image retrievals, this paper proposes a similarity retrieval method for trademarks using vector images. In the proposed method, an angle histogram which represents characteristics of an object is produced to every object in a vector image. And then, using features obtained from the histogram, similarity of objects between images is measured. Experimental results have shown that the proposed method could well evaluate similarity of each essential object in trademarks using vector-based images.

Quantum image segmentation algorithm can use its quantum mechanism to rapidly segment the objects in a quantum image. However, the existing quantum image segmentation algorithms can only segment static objects in the image and use more quantum resource(qubit). In this paper, a novel quantum segmentation algorithm based on background-difference method for NEQR image is proposed, which can segment dynamic objects in a static scene image by using fewer qubits. In addition, an efficient and feasible quantum absolute value subtractor is designed, which is an exponential improvement over the existing quantum absolute value subtractor. Then, a complete quantum circuit is designed to segment the NEQR image. For a ${2^n}$$\times$${2^n}$ image with gray-scale range of [0,$2^q$-1], the complexity of our algorithm is O($q$), which has an exponential improvement over the classical segmentation algorithm, and the complexity will not increase as the image's size increases. The experiment is conducted on IBM Q to show the feasibility of our algorithm in the noisy intermediate-scale quantum (NISQ) era.

Despite thermal imaging primarily used for nighttime surveillance, uniform temperature of object and background makes it difficult to acquire details in the scene being observed and thereby object detection. Further, thermal images collected over long distances degrade the spatial resolution of the acquired objects and so do the moving objects leading to noisy features. We present a computationally efficient object detection approach using Depthwise Deep Convolutional Neural Network (DDCNN) for detecting and classifying objects in nighttime images under low resolution. The Depthwise Convolution (DC) employed in the proposed approach minimises the network’s computational complexity resulting in the lowest number of training parameters (i.e., 3M) as compared to the other existing state-of-the-art methods such as FRCNN (52M), SSD (24M) and YOLO-v3 (61M) parameters. Further, by introducing novel Tversky and Intersection over Union (IoU) loss functions into the compact architectural design, we improve nighttime object detection accuracy. The validity of the proposed model is assessed on numerous datasets such as FLIR, KAIST, MS, and our internal dataset having multiple objects in each image. The experimental results from the proposed method indicate both quantitative and qualitative improvements over the recent state-of-the-art methods for nighttime imaging. The proposed approach achieves a mean Average Precision (mAP) of 52.39% and a highest individual object detection accuracy of 72.70% accuracy for cars in nigh-time situations suggesting applications in real-time use cases.

One new binarization method for gray images is presented in this paper by defining the decision rule, which puts not only much emphasis on local approach but also completely incorporates the shape-feature of the objects in the images, especially for the shape of the object in the gray images is known in advance. Experimental results demonstrate that our proposed method is generally capable of dealing with binarization challenges with high performance and is effective in solving problems of low contrast between the foreground and the background.

Rotation invariant HOG for object localization in web images

A recurrent curve matching classification method integrating within-object spectral variability and between-object spatial association

Purpose Image classification is a fundamental form of digital image processing in which pixels are labeled into one of the object classes present in the image. Multispectral image classification is a challenging task due to complexities associated with the images captured by satellites. Accurate image classification is highly essential in remote sensing applications. However, existing machine learning and deep learning–based classification methods could not provide desired accuracy. The purpose of this paper is to classify the objects in the satellite image with greater accuracy. Design/methodology/approach This paper proposes a deep learning-based automated method for classifying multispectral images. The central issue of this work is that data sets collected from public databases are first divided into a number of patches and their features are extracted. The features extracted from patches are then concatenated before a classification method is used to classify the objects in the image. Findings The performance of proposed modified velocity-based colliding bodies optimization method is compared with existing methods in terms of type-1 measures such as sensitivity, specificity, accuracy, net present value, F1 Score and Matthews correlation coefficient and type 2 measures such as false discovery rate and false positive rate. The statistical results obtained from the proposed method show better performance than existing methods. Originality/value In this work, multispectral image classification accuracy is improved with an optimization algorithm called modified velocity-based colliding bodies optimization.

In this paper, For the task of the depth map of a scene given a single RGB image. We present an estimation method using a deep learning model that incorporates size perspective (size constancy cues). By utilizing a size perspective, the proposed method aims to address the difficulty of depth estimation tasks which stems from the limited correlation between the information inherent to objects in RGB images (such as shape and color) and their corresponding depths. The proposed method consists of two deep learning models, a size perspective model and a depth estimation model, The size-perspective model plays a role like that of the size perspective and estimates approximate depths for each object in the image based on the size of the object's bounding box and its actual size. Based on these rough depth estimation (pre-depth estimation) results, A depth image representing through depths of each object (pre-depth image) is generated and this image is input with the RGB image into the depth estimation model. The pre-depth image is used as a hint for depth estimation and improves the performance of the depth estimation model. With the proposed method, it becomes possible to obtain depth inputs for the depth estimation model without using any devices other than a monocular camera be forehand. The proposed method contributes to the improvement in accuracy when there are objects present in the image that can be detected by the object detection model. In the experiments using an original indoor scene dataset, the proposed method demonstrated improvement in accuracy compared to the method without pre-depth images.

The article considers the tasks related to the prevention of emergency situations and the assessment and prediction of the consequences of natural disasters. It is shown that the effectiveness of these problems solution depends to a high degree on the efficiency of the analysis of the existing situation and the current state of the objects of surveillance in hazardous areas. At the same time, the most important source of information for decision-making are space images, primarily digital images, since they can be transmitted over high-speed radio-frequency transmission lines from the spacecraft to ground space information receiving points immediately after the survey. Scanner images obtained by optoelectronic scanning systems are among them and of special importance, since optoelectronic scanning systems ensure the registration of vast territories with high detail. During the ground images processing, two main tasks are solved, namely, the localization of surveillance objects in satellite imagery, connected with obtaining preliminary assessments and issuing target designations to ground forces and facilities for performing immediate operations aimed at eliminating existing threats, and detailed analysis of the situation and the current state of the surveillance objects with the clarification of previously issued target designations. In this case, the highest requirements for operational response are imposed on the solution of the first task. They are caused by the need to find the location of surveillance objects on the space imagery being processed as soon as possible. The known techniques of the localization of surveillance objects on space scanner images, namely, visual, photogrammetric (direct and iterative), correlation and approximation ones are considered. It is shown that on the one hand, in conditions of a huge amount of incoming for processing specific space information, and limited hardware and software resources of ground stations for processing space information, on the other hand, photogrammetric techniques for localizing surveillance objects on the survey route are the most acceptable ones. The known photogrammetric techniques are applied in the interactive mode and do not require large computational resources, since they are based on simple algebraic calculations in each iteration for a single point of an image. At the same time, the scope of the technique is critical to the duration of the survey route being processed, since along with the technological operations performed in the automatic mode (selection of image fragments) it contains the operations of the operator performing a visual assessment of the presence of surveillance objects in images. To eliminate the identified shortcomings, a technique for automatic localization of surveillance objects in space scanner images of vast territories by organizing their photogrammetric processing was proposed and considered. To test the efficiency of the technique and to study its effectiveness, an algorithm for constructing prototype scanner images on the surface of a common terrestrial ellipsoid has been developed.