Many satellite photos are distorted by noise, which lowers the visual quality and inevitably causes processing problems. To reduce this noise, a number of remedies have been put forth utilizing different strategies.
Image Credit: Andrei Armiagov/Shutterstock.com
Digital image processing is the computer analysis of digital images. Satellite images and other forms of images are represented electronically in digital images.
Prior to the computing process, image data can be improved through image processing. Image filtering is one of the pre-processing techniques that enables numerous digital manipulations of the images, removing noise, maintaining edges, and enhancing image quality. Images of the earth taken through a satellite orbiting it are referred to as satellite images. Satellite image processing is the process of removing haze, clouds, and sensor-induced flaws from these satellite photos and superimposing a 2D satellite image on a 3D model.
Applications of Satellite Images
Satellite images are used in many applications to acquire information about particular terrains and make informed decisions. For example, in agriculture, planning and controlling natural resource management requires land use mapping and monitoring. Satellite images are also widely used in emergency responses after a major hazard during rescue missions. Many other industrial applications, like geology, forestry, biodiversity preservation, regional planning, education, and warfare also use satellite images to optimize resource allocation and performance.
Satellite Image Noise
When raw satellite images are processed, environmental and systematic interference can appear in the final image. Any unwanted information that distorts an image is known as noise. Several sources of noise are visible in the image. The main mechanism through which noise emerges in digital images is the digital image acquisition process, which transforms an optical image into an electrical signal that is subsequently sampled. Depending on how the image is produced, noise can distort the image in a number of different ways.
It is necessary to employ appropriate filters to restrict or reduce most of the noise in satellite images since it can distort the image and make it difficult to comprehend. It increases the likelihood of a more accurate interpretation of the image content.
Typical noises sources that appear on a satellite image can be classified into one of the following types of noises:
- Random Variation Impulsive Noise -This kind of noise, also known as normal noise or Gaussian noise, appears at random as white intensity values on a digital image.
- Salt & Pepper Noise - This type of noise, which is frequently produced by noise image threshold, contains sporadic appearances of both black and white intensity excursions.
- Speckle Noise - Speckle noise is a common phenomenon that restricts the interpretation of optical coherence in remote sensing images if multiplicative noise is introduced to the image.
It is necessary to employ appropriate filters to restrict or reduce various noise effects from satellite images. Typical filters used for Satellite image noise reduction are:
- Gaussian Filter - Gaussian filters are made to minimize rise and fall times while providing no overshoot to step function inputs. In the 2D convolution operation, the Gaussian smoothing filter is employed to eliminate noise and blur from the image.
- Mean Filter - The Mean Filter is a straightforward linear filter that is simple to understand and use as a method of image smoothing.
- Standard Median Filter - If the spatial noise distribution in the image is not symmetrical inside the window, the median filter, which is a non-linear filter, modifies the mean value of image intensity. The variance of the intensities in the image is reduced by the median filter.
- Adaptive Median Filter - The Standard Median Filter's drawbacks are addressed by the Adaptive Median Filter. The Adaptive Median Filter's changeable window size around each pixel is the primary distinction between the two filters.
- Adaptive Wiener Filter - Based on the statistical properties of the picture included within the filter window, the adaptive Wiener Filter modifies its behavior. The performance of adaptive filters is typically better than that of non-adaptive counterparts. However, the extra filter complexity comes at the expense of enhanced performance.
Algorithmic Processing Methods
Large amounts of Earth observation (EO) data are now accessible online thanks to the current generation of open-access remote sensing satellites. A series of images received from the same region from the repeated orbits of remote sensing satellites can be combined into a time series to track change after the proper calibrations. Intricate underlying processes can be processed by detection methodologies using time series produced from EO satellite photos.
Depending on the application, a variety of solutions have been put forth to eliminate noise. Recent approaches have implemented AI-based techniques to analyze and mitigate noise contributions in satellite images. Some methods are listed below:
- Noise reductions based on convolutional neural networks.
- Principal Component Analysis and Data Masking for Noise Reduction
- Using multi-criteria decision-making technique, adaptive selection of edge detection algorithms to filter visual characteristics of satellite images.
- Adaptive weight algorithm methods.
- Creating time series clusters using self-organizing maps and using Bayesian inference to determine similarities between clusters.
All the noise reduction techniques mitigate the noise contributions in satellite images to various degrees effectively. Improvements to existing algorithms and new methods are continuously being developed in order to detect and suppress more complicated synthetic noise.
More from AZoOptics: The Future of Laser Communications in Space
References and Further Reading
Lorena A. Santos, Karine R. Ferreira, Gilberto Camara, Michelle C.A. Picoli, Rolf E. Simoes, Quality control and class noise reduction of satellite image time series, ISPRS Journal of Photogrammetry and Remote Sensing, Volume 177, 2021, Pages 75-88, ISSN 0924-2716, https://doi.org/10.1016/j.isprsjprs.2021.04.014.
Mr. Yogesh V. Kolhe, Dr. Yogendra Kumar Jain, 2013, Removal of Salt and Pepper Noise from Satellite Images, INTERNATIONAL JOURNAL OF ENGINEERING RESEARCH & TECHNOLOGY (IJERT) Volume 02, Issue 11 (November 2013). https://www.ijert.org/removal-of-salt-and-pepper-noise-from-satellite-images-2
Al-amri, Salem Saleh, Namdeo V. Kalyankar and Santosh D. Khamitkar. “A Comparative Study of Removal Noise from Remote Sensing Image.” ArXiv abs/1002.1148 (2010): https://www.semanticscholar.org/paper/A-Comparative-Study-of-Removal-Noise-from-Remote-Al-amri-Kalyankar/0f157e74c1d18ccda6ed11b1b8fa62cfa432a386
Disclaimer: The views expressed here are those of the author expressed in their private capacity and do not necessarily represent the views of AZoM.com Limited T/A AZoNetwork the owner and operator of this website. This disclaimer forms part of the Terms and conditions of use of this website.