A Learning-Based Resegmentation Method for Extraction of Buildings in Satellite Images

Makoveichuk

et al. 2023

EEJET

The object of this study is the high level of errors of the first and second kind in the segmentation of images of urbanized areas acquired from space optoelectronic surveillance systems. The method of image segmentation of urbanized areas implies two stages and, unlike known ones: – takes into account each channel of brightness of the color space of the original image; – at the first stage, an ant algorithm is used; – image segmentation at the first stage is reduced to the calculation of the objective function, the areas of movement of ants, and the concentration of pheromone on the routes of ant movement. – at the second stage, the brightness and geometric shape of the elements of objects are taken into account; – contours and geometric primitives are defined in the Hough parameter space; – the objects of interest of the urbanized area in the space of the original image are determined. An experimental study into the segmentation of images of urbanized terrain acquired from space optoelectronic observation systems was carried out based on the ant algorithm and the Hough algorithm. The quality of image segmentation of the urbanized area was assessed. It was found that the error of the first kind when using the improved method of segmentation is reduced by 2.75 %. The error of the second kind is reduced by 3.91 % when using the improved method of segmentation. This reduction became possible due to the use of an improved method of segmenting the image of an urbanized area by the ant algorithm at the first stage. Compared to Canny's algorithm, the error of the first kind decreased by 8.9 %, and the error of the second kind decreased by 11.0 %. Methods for segmenting images of urbanized areas acquired from space optoelectronic surveillance systems can be implemented in software and hardware systems of image processing

Section: Literature Review and Problem Statementmentioning

confidence: 99%

Section: Literature Review and Problem Statementmentioning

confidence: 99%

Section: Literature Review and Problem Statementmentioning

confidence: 99%

See 1 more Smart Citation

Development of a two-stage method for segmenting the color images of urban terrain acquired from space optic-electronic observation systems based on the ant algorithm and the hough algorithm

Makoveichuk

et al. 2023

EEJET

2014 International Conference on Control, Instrumentation, Communication and Computational Technologies (ICCICCT)

“…From the above computation, the QDA algorithm given by (6) with mean and (7) with covariance are approximated using the empirical mean and covariance of the data in ground-truth class i [14,15]. The classified set is then compared to the ground-truth pixels, respectively, and the percent of misclassified pixels is noted.…”

Section: Quadratic Discriminant Analysis Based Classifiermentioning

confidence: 99%

Land cover classification of multispectral satellite images using QDA classifier

Menaka

Suresh

Kumar³

2014

This paper presents a scheme for the classification of multispectral satellite images into multiple predefined land cover classes. The proposed approach results in a fully automatic classification model to assign each pixel in the image to a group of pixels based on reflectance or spectral similarity where each subset of group of pixels is called ground-truth data. The input image is preprocessed and applied to a classifier. The proposed supervised classifier incorporates both spectral and spatial information. We implement QDA Classifier (Quadratic Discriminant Analysis) based on spectral features. The classified image is then post processed using Probabilistic Label Relaxation algorithm for smoothening the output image which gives better results. The QDA Classifier uses statistical classification to separate measurements of two or more classes of objects or events by a quadric surface. It estimates the probability of each class across the spectral domain that it takes into account the correlations of the data set from the class centroid. An experiment on multispectral satellite images shows the accuracy of the method.

“…The extraction and analysis of earth surface features through high resolution remote sensing (HRRS) images has received extensive research, such as the buildings extraction [1][2][3], vegetation detection [4][5][6], urban expansion analysis [7][8][9] and detection of land cover changes [10]. However, there is a key issue that cannot be ignored: ensuring the security of HRRS image is the basic prerequisite for using HRRS images.…”

Section: Introductionmentioning

confidence: 99%

Semi-U-Net: A Lightweight Deep Neural Network for Subject-Sensitive Hashing of HRRS Images

Ding

Su²,

et al. 2021

IEEE Access

As a special case of perceptual hashing algorithm, subject-sensitive hashing can realize "subject-biased" integrity authentication of high resolution remote sensing (HRRS) images, which overcomes the deficiencies of existing integrity authentication technologies. However, the existing deep neural network for subject-sensitive hashing have disadvantages such as high model complexity and low computational efficiency. In this paper, we propose an efficient and lightweight deep neural network named Semi-U-net to achieve efficient subject-sensitive hashing. The proposed Semi-U-net realizes the lightweight of the network from three aspects: First, considering the general process of perceptual hashing, it adopts a semi-u-shaped structure, which simplify the model structure and prevent the model from extracting too much redundant information to enhance the robustness of the algorithm; Second, the number of model parameters and the computational cost are significantly reduced by using deep separable convolution in the entire asymmetric network; Third, the number of model parameters is further compressed by using the dropout layer several times. The experimental results show that the size of our Semi-U-Net model is only 5.38M, which is only 1/27 of MUM-net and 1/15 of MultiResUnet. The speed of the Semi-U-Net based subject-sensitive hashing algorithm is 88.6 FPS, which is 2.89 times faster than MultiResUnet based algorithm and 2.1 times faster than MUM-net Based Algorithm.