Pre processing algorithm of multispectral image strips of an aerial filter array

DOI:	10.11809/bqzbgcxb2023.04.037
Keywords:	filter array; multi spectral; strip; greyscale; spectral mixing
Abstract:	Aiming at the problem of inaccurate cropping of the effective region of the strips due to low strip overlap and spectral mixing at the edges of the strips in multispectral images of an aerial filter array, this paper proposes an pre processing algorithm of multispectral image strips of an aerial filter array. The greyscale extreme points between strip images are calculated successively according to the greyscale mean value of row pixels of multispectral images of an aerial filter array. The intersection between the minimum mean greyscale value of the row pixels in the middle region of the strip images and the grey change curve of its row pixels is taken as the strip cut off point, and the difference between the row direction coordinates corresponding to the extreme greyscale point and the strip cut off point is figured out to obtain the edge spectrum mixing range of the strip images. According to the maximum edge spectral mixing width of each band image and the row direction coordinates of the adjacent grey extreme points, the vertex coordinates of the effective region of each band are successively calculated, and the single band strip images are obtained by image cropping. The theoretical analysis and experimental results show that this method can accurately extract the effective regions of each strip images, preserve the overlapping region pixels to the maximum extent, and, with no spectral mixing, the effective areas of each cropped strip have a consistent change in pixel greyscale.
Issue:	Vol. 44 No. 4 (2023)
Published:	2023-04-28
PDF HTML