Удосконалення методу оперативного визначення ділянки фантомізаціі на зображенні з космічної системи радіолокаційного спостереження

This work examines the process that determines a phantom area in an image acquired from a space radar observation system. The principal hypothesis of the study assumed that improving a method for defining the phantom area could reduce image processing errors of the first and second kind. The operational method for determining the phantom area in an image acquired from a space radar observation system has been improved; in it, in contrast to known ones, – a radar image is represented as a two-dimensional array of pixels whose intensity is determined by the amplitude of the radar signal in grayscale; – the influence of speckle noise is minimized using convolution with a Gaussian filter; – the image histogram is aligned to increase contrast; – the boundaries in the image are selected using a gradient operator; – the area with the selected boundaries with objects of interest is defined as the phantom area in the image acquired from a space radar observation system. An experimental study was conducted on the operational determination of the phantom area in an image from a space radar surveillance system. Phantom areas on which objects of interest are located are highlighted in an image from a space radar surveillance system. At the final stage of the improved method for boundary selection, the Sobel, Prewitt, and Roberts operators are considered. The choice of the Roberts operator at the final stage for boundary selection allowed for the following: – a decrease in processing errors of the first kind: by 2.64% compared to the Sobel operator; by 5.66% compared to the Prewitt operator; – a reduction of processing errors of the second kind: by 2.4% compared to the Sobel operator; by 4.26% compared to the Prewitt operator