Abstract
Abstract. The quality of satellite images has always been of particular importance in remote sensing. Signals received from satellite sensors include some signals other than those of target signal that may be classified totally as the atmospheric effect and the sensor induced noise. Separating non-target signals and attempting in removing them from images is essential. One method for measuring and removing non-target signals is that of atmospheric correction by Dark Object Subtraction (DOS). This method is based on the sensor’s output for the targets that should have almost zero reflectance in a given band. Next, the obtained value will be deducted from the remaining pixels values; regardless of the type of the sensors. Each Charge-Coupled Device (CCD) has its own noise behavior; therefore, the amount deducted values from each pixel can be different for each CCD unit and type. Among the various noises of the CCD and their related electronic circuits, dark current noise, non-uniform pixels noise and read noise were selected to be studied in this paper. The data were obtained from multispectral sensor images of IKONOS. This sensor can provide images in two forms of Panchromatic (PAN) and Multispectral (MS). The results of this study showed that the amount of dark object pixels and the total amount of CCD noises in each band are different. Separation of the noises introduced in this paper from the amount of dark object pixel values can result in an upgraded method for image atmosphere corrections.
Highlights
The world's first commercial satellite was launched on September 24, 1999 with the name of IKONOS
Signals received from satellite sensors; often include some signals other than target signals, the group of which are defined as non-target signals
The nontarget and the sensor noise which mostly includes Charge-Coupled Device (CCD) noises are removed from the image
Summary
The world's first commercial satellite was launched on September 24, 1999 with the name of IKONOS. The amount of energy emitted (radiance or reflectance) from some earth targets in certain wavelengths, is close to zero (Pagnutti, 2003) In this case, the energy received by the sensor comes from non-target signals along with the noise of imaging system used. The dark pixels are identified and deducted from the surrounding pixels By this way, the nontarget and the sensor noise which mostly includes CCD noises are removed from the image. The specified noises for each band are calculated separately Each of these noises has different effects on image pixel values. Separating these noises from the dark object ones; makes it possible to predict electronic noise via changing the sensor, and to obtain a constant value for the sensor noise values. 0.45-0.53, (blue) 0.52-0.61, (green) 0.64-0.72, (red) 0.76-0.86, (NIR) ADPCM, 2.5 bits/pixel 11 bits
Talk to us
Join us for a 30 min session where you can share your feedback and ask us any queries you have