Destriping Of TDI-CCD Remote Sensing Image

CCD remote sensing camera which uses CCD as an image sensor is an important application in the field of image sensor. TDI-CCD data are polluted by obvious and periodic striping noise over the entire image because the photoelectricity response characteristic of intrinsic semiconductors and the constraints of structural design. If not removed properly, the striping noise can reduce the detection sensitivity of TDI-CCD measuring system and decrease the dynamic range of work. Some effective measures should be taken to eliminate the striping noise.Based on the characteristic of striping noise in multi-channel TDI-CCD remote sensing images, four new destriping noise techniques are presented in this letter.1.A destriping noise technique which combines FFT with wavelet transform to search the frequency point of the striping noise is presented. The method has used the diagram of the discrete cumulative distribution of the image frequency spectrum and the wavelet transform to detect the singular points. The improved notch filter is selected to eliminate the striping noise.2.Based on the advantage of singular points detection of the wavelet transform and the curve of the column summation, we can obtain the positions of the singular points which are the position of the striping noise. Then use the appropriate interpolation method to eliminate the striping noise.3.A new destriping noise technique for the improved multi-threshold method using wavelet transform is applied to remote sensing imagery. Have used the lifting wavelet decomposition algorithm, each scale has its corresponding threshold and the threshold is determined by corresponding wavelet coefficients in every scale. This method can overcome the shortcoming of the single threshold and has achieved good result.4.A new destriping technique for the improved threshold function using wavelet transform is also presented in this letter. It can overcome the shortcoming of the hard threshold function which is of discontinuity and soft threshold function which is of warp simultaneity.Experimental results show that the proposed methods can effectively eliminate the striping noise in multi-channel TDI-CCD images while preserve the characteristic of the original images.Then the methods can achieve good destriping result, so the several methods are of feasibility.