16-BIT TO 8-BIT CONVERSION IN REMOTE SENSING IMAGES BY USING IMAGE CONTENT

Emre BAŞESKİ

Abstract:

16-bit to 8-bit image conversion, with minimal visual loss, is an important problem for research areas such as remote sensing and medical imaging. When the color tones obtained during the conversion are too light or too dark, some visual patterns in the image can be lost. In this study, an algorithm that automatically converts a 16-bit image to 8-bit without user intervention is presented. The algorithm is based on an artificial neural network that can produce images with well tuned brightness and contrast values by using a model that was learned from satellite images with different content. The proposed algorithm takes into account the visual content of the image while reducing the color depth. Therefore, it can produce more natural looking and visually more plausible results than the algorithms that only uses color distribution for 8-bit conversion.
 

Detaylı bilgi için lütfen tıklayınız.