RESTORING THE INFORMATION CHARACTERISTICS OF OBJECTS ON MULTISPECTRAL DIGITAL REMOTE SENSING IMAGES

Authors

  • V.M. KORCHYNSKYI

DOI:

https://doi.org/10.32782/KNTU2618-0340/2021.4.2.1.14

Keywords:

multispectral image, intrinsic brightness, homoform transformation, orthogonalization, dimension reduction, ideal device, energy information entropy, structural similarity index

Abstract

The method of determining the reflectivity (own brightness) of material objects by their isplanatical multispectral images, recorded in short-wave radiation ranges - the carrier of species information - is proposed. The brightness distributions of these images are presented in the form of a roll-up of the object's own brightness and the hardware function of the image sensor. The proposed method is based on the analysis of spectral representations of the brightness of images in the area of spatial frequencies, the transformation into which is carried out by a two-dimensional transformation of Fourier's pronounced bundle. As a model of transmission characteristics of image sensors are considered low-frequency spatial filters, characteristic of sensors in the form of focusing systems, with cut-off frequencies, inversely proportional to the wavelength of radiation - the carrier of species information. A two-dimensional array in the form of a concatenation of vectors representing the distributions of the Fourier spectra of the brightness functions of individual spectral channels is used as an information carrier for representations of digital multispectral images in the region of spatial frequencies. The restoration of the intrinsic brightness of the depicted objects was carried out in terms of the general theory of linear measurements and carried out by constructing a linear operator that implements the concept of an "ideal device" compensating for the influence of the hardware function of the image sensor. The actual reconstruction of the spatial distributions of the intrinsic brightness of objects is performed by the inverse two-dimensional Fourier transform in the region of spatial frequencies. An entropic interpretation of the proposed method is proposed as a method for filtering information components of digital multispectral images that are insignificant for their thematic analysis, provided that the spatial spectra of the recorded images and frequency transfer characteristics of the sensor of the view information are statistically independent, and the corresponding quantitative estimates are given. It was found that the proposed method provides a high level of preservation of the structural similarity of the brightness distribution functions of the original and reconstructed images of spectral channels.

References

Шовенгердт Р.А. Дистанционное зондирование. Модели и методы обработки изображений. Москва: Техносфера, 2010. 560 с.

Sun W., Xu G., Gong P., Liang S. Fractal Analysis of Remotely Sensed Images: A Review of Methods and Applications. International Journal of remote Sensing. 2016. Vol. 47. № 2. P. 4963–4990.

Залманзон Л.А. Преобразования Фурье, Уолша, Хаара и их применение в управлении, связи и других областях. Москва: Наука, 2001. 496 с.

Смит С. Цифровая обработка сигналов. Практическое руководство для инженеров и научных работников. Москва: Додэка-ХХІ, 2021. 720 с.

Пытьев Ю.П. Задачи редукции в экспериментальных исследованиях. Математический сборник. 1993. Том 120. № 2. С. 145–158.

Исимару А. Распространение и рассеяние волн в случайно-неоднородных средах. Том 2. Многократное рассеяние / турбулентность / шероховатые поверхности и дистанционное зондирование. Москва: Наука, 1989. 318 с.

Wang Z., Bovik A.K, Sheikh H.R., Simoncelli E.R. Image Quality Assessment: From Error Visibility to Structural Similarity. IEEE Transactions on Image Processing. 2004. Vol. 13. №, 4. P. 600–612.

Кудряшов Б.Д. Теория информации. Санкт-Петербург: ПИТЕР, 2009. 320 с.

Published

2023-04-14