Research On Spectral Correction And Color Interpolation Of Color Nightlight Remote Sensing Image

As one of the main development directions in the field of remote sensing,nightlight remote sensing has attracted more and more attention in the field of natural science and social economy in recent years.Nightlight remote sensing mainly uses remote sensing satellites to capture visible light radiation sources such as town lights,fishing boat lights and fire spots.Compared with ordinary remote sensing satellite images,nightlight remote sensing images can more intuitively show the differences of human activities,so nightlight remote sensing is widely used in social and economic fields.In order to better retain the real spectral radiation information in the process of remote sensing imaging,this thesis mainly researches the out of band spectral response correction and color interpolation algorithm of nightlight remote sensing satellite.First,for the radiation mixing between different spectral channels in the remote sensing imaging process,namely out of band response,this thesis proposes a method of out-of-band spectral response correction of nightlight remote sensing satellites.According to the spectral response model of nightlight remote sensing camera,the out of band response correction model is established by analyzing the causes of out of band spectral response in the imaging process.The out of band response correction coefficient is obtained by calculating the spectral response function of each spectral segment of the nightlight remote sensing camera and the standard light source with different spectral characteristics,and then the out of band response correction of Bayer format image is completed.Secondly,because the image after the out of band response correction is still a single-channel Bayer format image,in order to obtain a complete color image and minimize the impact of the color interpolation process on the spectral radiation accuracy,this thesis proposes an improved color interpolation algorithm.The algorithm is mainly divided into three steps: use the edge-sensitive interpolation algorithm to make the initial pixel estimation value of the G channel,establish the color difference plane between the initial estimation value of the G channel and the different channels,restore the pixel estimated values of other channels,and calculate Complete the update and refinement of the G channel pixels;in order to further optimize the G channel pixels,the color difference plane between the G channel and other channels is reconstructed by the expansion of the reference range and the combination of directionality,and the weight values of different directions are combined with the color difference plane to establish The new color difference relationship is added to the original R and B channels to restore the complete G channel plane;the R and B channel pixels are reconstructed using the guided filtering method.Finally,this thesis uses remote sensing satellites to shoot the target light source in orbit and retrieve the radiance information,and compares it with the measured value on the ground.Through the analysis and comparison results,the proposed out-of-band spectral response correction method and color interpolation algorithm are experimentally verified.Experimental results show that the out of band response correction method and color interpolation algorithm proposed in this thesis can better retain the spectral radiance information of the target light source,and verify that the method in this thesis can effectively improve the radiation measurement accuracy of color nightlight remote sensing cameras.