}I hope this will help you the median cut is quite a nice algorithm, it's quite fast and it's results are very nice. Else: Red has greater part else if (mm.Rmax - mm.Rmin > mm.Gmax - mm.Gmin)įor (iCount = 0, R = mm.Rmin R = iBoxCount / 2)įor (iCount = 0, G = mm.Gmin G = iBoxCount / 2) If next blue step is maximum, also finish if (B = mm.Bmax - 1)īoxCut (piCount, iCount, mmNew, iRes, iLevel + 1, pboxes, piEntry) īoxCut (piCount, iBoxCount - iCount, mmNew, iRes, iLevel + 1, pboxes, piEntry) Loob blue parts for (iCount = 0, B = mm.Bmin B = iBoxCount / 2) Else: if blue has greater part, divide else if ((mm.Bmax - mm.Bmin > mm.Rmax - mm.Rmin) & PalFindAverageColor (piCount, mm, iRes, &pboxes.rgbBoxAv) If recursion reached eigth level, or cube has only one pixel left, // get common color and save with count of pixels of this color if (iLevel = 8 || (mm.Rmin = mm.Rmax & If cube contains no pixels, break if (iBoxCount = 0) Return pbox1->iBoxCount - pbox2->iBoxCount ISDLLEXPORT int boxCompare ( const DPALBOXES * pbox1, }Message too long see next message for the other parts. Plp->palPalEntry.peBlue = ĭata->status (dilGetMessage(DIBMSG_COLORTABLECREATE)) Plp = reinterpret_cast(malloc ( sizeof(LOGPALETTE) + (iEntry - 1) * sizeof(PALETTEENTRY))) ĭata->status (dilGetMessage(DIBMSG_COLORTABLECOPY)) įor (i = 0 i palPalEntry.peRed = Qsort (boxes, iEntry, sizeof(DPALBOXES), reinterpret_cast(boxCompare)) Divide cube // colors are reduced to 256, one for each cube // iEntry counts non-used non-empty cubesĭata->status (dilGetMessage (DIBMSG_MEDIANCUTBOXCUT)) īoxCut (piCount, iTotCount, mm, nRes, 0, boxes, &iEntry) ĭata->status (dilGetMessage(DIBMSG_MEDIANCUTQSORT)) IDim = 1 status (dilGetMessage(DIBMSG_MEDIANCUTBOX)) PalAccumColorCounts (hDib, piCount, nRes) If (NULL = (piCount = reinterpret_cast(calloc (iArraySize, sizeof( int)))))ĭata->status (dilGetMessage (DIBMSG_ACCUMCOLORS)) IArraySize = 1 status (dilGetMessage (DIBMSG_ALLOCMEMORY)) Check for palette image if (dibGetBitCount (hDib) 8)
Int i, iArraySize, iCount, R, G, B, iTotCount, iDim, iEntry = 0 ISDLLEXPORT HPALETTE palCreateMedianCut (HDIB hDib, Print "Converting " + rname + " to float"Īrcpy.RasterToFloat_conversion(os.path.join(tmpDIR,temp + "c3"), os.path.join(tmpDIR, fltname + "_c3.flt"))Īrcpy.RasterToFloat_conversion(os.path.join(tmpDIR,temp + "c5"), os.path.join(tmpDIR, fltname + "_c5.flt"))Īrcpy.CopyRaster_management(os.path.join(tmpDIR, fltname + "_c5.flt"), os.path.join(tmpDIR, \ Print "Converting " + rname + " to 8 Bit Image"Īrcpy.CopyRaster_management("tempc5", "band5.tif", "", "", "", "NONE", "ColormapToRGB", "8_BIT_UNSIGNED")Īrcpy.CopyRaster_management("tempc3", "band3.tif", "", "", "", "NONE", "ColormapToRGB", "8_BIT_UNSIGNED")Īrcpy.CopyRaster_management("tempc2", "band2.tif", "", "", "", "NONE", "ColormapToRGB", "8_BIT_UNSIGNED")Īrcpy.CompositeBands_management("band5.tif band3.tif band2.tif", os.path.join(bitDIR,rname)) Anybody see something that I'm missing?Īrcpy.CopyRaster_management(raster, os.path.join(tmpDIR,temp), "", "", "", "NONE", "NONE", "32_BIT_UNSIGNED") NDVI routine runs fine as it continues to work fro the GRID files. Problem is happening in the coversion to 8 bit. I checked the GRID files and they are correct. The problem that I am having is that each CopyRaster_management routine I run to convert the bands to 8 bit tif's produces a 8 bit file with all the pixels having the same value. I then composite them in the order I want into a single. I take the 3 raster bands (ESRI GRIDS) that I want to composite and convert them to 8 bit tif files. The next part is what is breaking down for me (highlighted in dark red). I start by converting the geo-tif to a ESRI GRID. I'm doing this in conjunction with calculating an NDVI for each image as well.
8 bit color converter 32 bit#
Trying to put together a python script for converting my 5-band, 32 bit unsigned to image to a 3-band, 8 bit unsigned image.
0 kommentar(er)
