3.2. Image Acquisition and Preprocessing

Radiometric Correction

• Correction of sensor irregularities and atmospheric interference.
• Converts raw Digital Numbers (DN) into calibrated reflectance values.

Radiometric correction is a process used to address variations in the brightness of images that may arise due to sensor irregularities or atmospheric conditions. Sensors on satellites capture sunlight reflecting off the earth's surface, but they can introduce errors. By applying this correction technique, raw images (in Digital Numbers) are adjusted to reflect light measured in a way that can be consistently interpreted as reflectance values. This standardized conversion allows for more accurate comparisons and analyses of satellite images across different times and locations.

Think of radiometric correction like adjusting the colors on a photo after taking it with your phone. Sometimes the lighting isn't perfect, so you adjust the brightness and contrast to make the image look more natural. Similarly, radiometric correction 'tweaks' satellite images to ensure they are consistent and accurate, allowing researchers to analyze changes in land cover over time effectively.

Geometric Correction

• Aligns satellite images to real-world coordinates.
• Involves Ground Control Points (GCPs) and resampling techniques such as nearest neighbor, bilinear interpolation, and cubic convolution.

Geometric correction ensures that the satellite images align accurately with geographic coordinates on the Earth’s surface. This is essential because satellite images can be distorted due to the angle of the satellite, the curvature of the Earth, or sensor-related issues. Ground Control Points (GCPs) — known geographic locations that can be identified on both the satellite images and map data — are used to accurately place the image. Resampling techniques, like nearest neighbor or bilinear interpolation, help in precisely mapping these images to their real-world locations without losing important detail or introducing distortion.

Imagine you’re putting together a puzzle where the pieces are slightly warped. You need to stretch or adjust some pieces to fit correctly in their spots. Similarly, geometric correction is like reshaping satellite images so that they correctly align with their real-world counterparts, just like ensuring each puzzle piece fits in the right place.

Atmospheric Correction

• Eliminates effects of atmospheric scattering and absorption.
• Common methods: Dark Object Subtraction (DOS), Fast Line-of-sight Atmospheric Analysis of Spectral Hypercubes (FLAASH).

Atmospheric correction deals with the distortions caused by the atmosphere when satellite images capture the earth's surface. Factors like air pollution, humidity, and atmospheric gases can scatter light and affect how we perceive colors in images. Techniques such as Dark Object Subtraction (DOS) involve identifying and removing the influence of these atmospheric effects on specific dark parts of an image, while methods like FLAASH use more complex models to account for varying atmospheric conditions across the image. This correction leads to more reliable and accurate interpretations of surface features.

Imagine viewing a beautiful sunset while wearing glasses smeared with smog. The colors look dulled and distorted because of the obstructions in your view. When the smog is cleared, the brilliance of the colors returns. Atmospheric correction is similar; it clears up satellite images so they reflect true colors and details of the earth's surface, enhancing the quality of the data for analysis.

Noise Reduction

• Filters used to remove random or systematic noise.
• Spatial filtering techniques like Median, Gaussian, and Low-pass filters are employed.

Noise reduction is essential in satellite imagery since images can contain random noise (unwanted signals or variations) that obscure important details. Various filtering techniques are employed for this purpose. For example, a Median filter replaces each pixel's value with the median value of its neighborhood, effectively removing outliers while preserving edges. Gaussian filters apply a bell-shaped curve to average pixel values, helping to smooth the image. Low-pass filters do similar work by allowing low-frequency signals (smooth areas) and blocking high-frequency signals (noise) to enhance the quality of the imagery.

Think about recording a song in a noisy room. The background noise makes it hard to hear the singer's voice clearly. You would want to use noise reduction techniques in audio editing to isolate the voice and remove the disturbances. In satellite image preprocessing, noise reduction techniques help clear up the image, making the significant features of the landscape easier to identify and analyze.

--