Hosuk:LabNotes/2014-7-23
Jump to navigation
Jump to search
Deconvolution and PSF images - Update2
Better PSF Functions
- PSF_20x_4096x4096_best1st_256x256.tif
- PSF Richards and Wolf_63x_Sqrt_Stack21_z10.tif
- PSF Richards and Wolf_63xZoom4_Sqrt_Stack21_z10.tif
- 20x PSF is the same as previous one.
- PSF for 63x and 63x Zoom=4 are used 'Richard and Wolf' instead of 'Gibson and Lani'.
- Choose middle image (11th) after creating 21 stacks, which means use the smallest size of intensity distribution
Procedure
- deconvblind function can perform deconvolution with uint16 class images.
- deconvoluted result need to be double, normalized [0 1] for PISA run or other post rolony detection algorithm
For 20x, 4096x4096
- deconvblind Image --> medfilt2
- img_raw = imread(ImgFileName, 'tif');
- PSFimg = imread(PSFFileName, 'tif');
- [img_decv, PSF] = deconvblind(img_raw, PSFimg);
- img_flt = medfilt2(img_decv, [3 3]);
- hi = prctile( double( reshape(img_flt, 1, xsize*ysize) ), 99.9);
- img_decv_double = double(img_flt)/hi;
For 63x and 63x Zoom=4, 2048x2048
- deconvblind Image --> mat2gray
- img_raw = imread(ImgFileName, 'tif');
- PSFimg = imread(PSFFileName, 'tif');
- [img_decv, PSF] = deconvblind(img_raw, PSFimg);
- img_decv_double = mat2gray(img_decv);