The ability to correct haze in images is vital in creating an accurate, visually pleasing image mosaic. Haze not only obstructs valuable information about any given pixel, it also creates a visually displeasing effect in the image, often contributing to errors in color balancing in the final mosaic.
The process of detecting and removing haze begins with analyzing the haze statistics and associated haze, water, and cloud masks to evaluate and remove the additive effects of haze for a given pixel in multispectral imagery. The process also depends heavily on image statistics calculated over the area of the imagery and is based on pixels that are considered not hazy (clear), and those that are considered hazy.
A reasonable number of clear pixels are required to determine the correction to apply to the scene. For example, when the image is completely covered in haze, the algorithm cannot properly correct for haze, because there are no "good" pixels to compare against. This is an important concept to keep in mind when evaluating the results produced by the Cloud Detection and Haze Removal module.
© PCI Geomatics Enterprises, Inc.®, 2026. All rights reserved.