; $Id: search3d.pro,v 1.12 2001/01/15 22:28:11 scottm Exp $ ; Copyright (c) 1992-2001, Research Systems, Inc. All rights reserved. ; Unauthorized reproduction prohibited. ; ;+ ; NAME: ; SEARCH3D ; ; PURPOSE: ; This function finds "objects" or regions of similar data ; values within a 3-D array of data. Given a starting location ; and a range of values to search for, SEARCH3D will find all ; the cells within the volume that are within the specified range ; of values, and have some path of connectivity through these cells ; to the starting location. In addition to searching for cells ; within a global range of data values, SEARCH3D can also search ; for adjacent cells whose values deviate from their neighbors within ; specified tolerances. See the procedure "SEARCH2D" for the ; two dimensional case. ; ; This function returns a list of the array subscripts that define ; the selected object or region. ; ; CATEGORY: ; Data subsetting. ; Volume manipulation. ; ; CALLING SEQUENCE: ; Region = SEARCH3D(Array, Xpos, Ypos, Zpos, Min_val, Max_val) ; ; INPUTS: ; Array: The 3-D volume of data to search. ; Data type : Any 3-D array except string or structure. ; Xpos: The X coordinate (first subscript into the 3-D Array) ; of the starting cell for the search. ; Data type : Long. ; Ypos: The Y coordinate (second subscript into the 3-D Array) ; of the starting cell for the search. ; Data type : Long. ; Zpos: The Z coordinate (third subscript into the 3-D Array) ; of the starting cell for the search. ; Data type : Long. ; Min_val: The minimum data value to search for. All cells that ; are connected to the starting cell, and have a value ; greater than or equal to Min_val and less that or equal ; to Max_val, will be considered part of the "object". ; Max_val: The maximum data value to search for. ; ; KEYWORD PARAMETERS: ; DECREASE: If the DECREASE or INCREASE keywords are specified, ; then SEARCH3D creates an internal copy of Array. ; This internal copy is then processed to enhance the ; object edges by using an algorithm similar to the ; "SOBEL" edge enhancement process (in 3-D). Any ; adjacent cells will be found if their corresponding ; data value in the edge enhanced array is greater ; than DECREASE and less than INCREASE. In any case, ; the adjacent cells will NEVER be selected if their ; data value is not between Min_val and Max_val. ; The default is 0.0 if INCREASE is specified. ; Otherwise, the default is no edge checking. ; Data type : Int or Float (usually less than zero). ; INCREASE: The maximum value in the edge enhanced array for ; a cell to be considered part of the selected object. ; Some savings in execution time and memory usage result ; when DECREASE and INCREASE are omitted. ; See DECREASE above. ; The default is 0.0 if DECREASE is specified. ; Otherwise, the default is no edge checking. ; Data type : Int or Float (usually greater than zero). ; LPF_BAND: This keyword indicates what (if any) Low Pass Filtering ; is performed on the edge enhanced array before the ; search begins. If LPF_BAND is set to 3 or higher ; then the edge enhanced array will be smoothed using ; LPF_BAND as the width of the smoothing window. ; If LPF_BAND is less than 3 then no smoothing is ; performed. This keyword only has effect when the ; DECREASE or INCREASE keywords are supplied. ; See DECREASE above. ; The default is zero (no smoothing). ; Data type : Int. ; DIAGONAL: Normally, cells are considered adjacent only when ; cubes surrounding the cells share a common face. ; If a non-zero value is passed to DIAGONAL then ; SEARCH3D will also locate cells meeting the search ; criteria whose surrounding cubes share a common ; edge or corner. Specifying diagonal search mode ; requires more memory and execution time. ; The default is no diagonal searching. ; Data type : int ; ; OUTPUTS: ; This function returns a list of the indices into the 3-D array ; that are part of the located object or region. This list is ; returned as a LONARR(n) where n is the number of cells found. ; ; If the returned array of indices is called Region, and the ; size of the 3-D volume of data is size_x by size_y by size_z, ; then the actual X, Y, and Z indices can be obtained by using ; the following algorithm : ; ; index_z = Region / (size_x * size_y) ; index_y = (Region - (index_z * size_x * size_y)) / size_x ; index_x = (Region - (index_z * size_x * size_y)) - (index_y * size_x) ; ; The object within the 3-D Array could then be subscripted as : ; ; Array(Region) ; OR ; Array(index_x, index_y, index_z) ; ; EXAMPLE: ; Find all the indices corresponding to an object contained in a ; 3-D volume of data. ; ; ; Create some data. ; vol = RANDOMU(s, 40, 40, 40) ; vol(3:13, 1:15, 17:33) = 1.3 ; vol(15:25, 5:25, 15:25) = 0.2 ; vol(5:30,17:38,7:28) = 1.3 ; vol(9:23, 16:27, 7:33) = 1.5 ; ; ; Search for an object starting at (6, 22, 16) whose data values ; ; are between (1.2) and (1.4).. ; Region = SEARCH3D(vol, 6, 22, 16, 1.2, 1.4, /DIAGONAL) ; ; ; Scale the background cells into the range 0 to 127. ; vol = BYTSCL(vol, TOP=127B) ; ; ; Highlight the object region by setting it to 255. ; vol(Region) = 255B ; ; ; Set up a 3-D view. ; Window, 0, Xsize=640, Ysize=512, Retain=2 ; Create_View, Xmax=39, Ymax=39, Zmax=39, ax=(-30), az=30, zoom=0.8 ; ; ; Display the volume with the highlighted object in it. ; TVSCL, PROJECT_VOL(vol, 64, 64, 40, Depth_Q=0.4) ; ; MODIFICATION HISTORY: ; Written by: Daniel Carr. Thu Sep 3 17:36:04 MDT 1992 ;- FUNCTION Search3d, array, xpos, ypos, zpos, min_val, max_val, $ Decrease=decrease, Increase=increase, $ Lpf_band=smooth_band, Diagonal=diagonal ; *** Test inputs ON_ERROR, 2 size_array = Size(array) IF (size_array[0] NE 3L) THEN BEGIN MESSAGE, 'Array must have three dimensions' ENDIF x_size = size_array[1] y_size = size_array[2] z_size = size_array[3] xpos = Long(xpos[0]) ypos = Long(ypos[0]) zpos = Long(zpos[0]) IF (xpos LT 0L) THEN BEGIN MESSAGE, 'Xpos must be >= 0' ENDIF IF (xpos GE x_size) THEN BEGIN MESSAGE, 'Xpos must be < array size' ENDIF IF (ypos LT 0L) THEN BEGIN MESSAGE, 'Ypos must be >= 0' ENDIF IF (ypos GE y_size) THEN BEGIN MESSAGE, 'Ypos must be < array size' ENDIF IF (zpos LT 0L) THEN BEGIN MESSAGE, 'Zpos must be >= 0' ENDIF IF (zpos GE z_size) THEN BEGIN MESSAGE, 'Zpos must be < array size' ENDIF min_val = min_val[0] max_val = max_val[0] IF (max_val LT min_val) THEN BEGIN MESSAGE, 'Max value must be >= min value' ENDIF start_val = array[xpos,ypos,zpos] IF ((start_val LT min_val) OR (start_val GT max_val)) THEN BEGIN MESSAGE, 'Value of array at (xpos,ypos,zpos) must be >= min_val and <= max_val' ENDIF dec = 0.0 inc = 0.0 range = 0B IF (N_Elements(decrease) GT 0L) THEN BEGIN dec = Float(decrease[0]) range = 1B ENDIF IF (N_Elements(increase) GT 0L) THEN BEGIN inc = Float(increase[0]) range = 1B ENDIF sb = 0 IF (N_Elements(smooth_band) GT 0L) THEN sb = Fix(smooth_band[0]) IF ((sb GE x_size) OR (sb GE y_size)) THEN BEGIN MESSAGE, 'Smooth band must be < size of array' ENDIF diag = 0B IF (N_Elements(diagonal) GT 0L) THEN diag = Byte(diagonal[0]) IF (range) THEN BEGIN ; *** Calculate the edge enhanced array IF (diag) THEN BEGIN diff_array = Float(array) diff_array = diff_array < $ (diff_array - Shift(array, 0, 1, 0)) < $ (diff_array - Shift(array, 1, 1, 0)) < $ (diff_array - Shift(array, 1, 0, 0)) < $ (diff_array - Shift(array, 1, -1, 0)) < $ (diff_array - Shift(array, 0, -1, 0)) < $ (diff_array - Shift(array, -1, -1, 0)) < $ (diff_array - Shift(array, -1, 0, 0)) < $ (diff_array - Shift(array, -1, 1, 0)) < $ (diff_array - Shift(array, 0, 1, 1)) < $ (diff_array - Shift(array, 1, 1, 1)) < $ (diff_array - Shift(array, 1, 0, 1)) < $ (diff_array - Shift(array, 1, -1, 1)) < $ (diff_array - Shift(array, 0, -1, 1)) < $ (diff_array - Shift(array, -1, -1, 1)) < $ (diff_array - Shift(array, -1, 0, 1)) < $ (diff_array - Shift(array, -1, 1, 1)) < $ (diff_array - Shift(array, 0, 1, -1)) < $ (diff_array - Shift(array, 1, 1, -1)) < $ (diff_array - Shift(array, 1, 0, -1)) < $ (diff_array - Shift(array, 1, -1, -1)) < $ (diff_array - Shift(array, 0, -1, -1)) < $ (diff_array - Shift(array, -1, -1, -1)) < $ (diff_array - Shift(array, -1, 0, -1)) < $ (diff_array - Shift(array, -1, 1, -1)) < $ (diff_array - Shift(array, 0, 0, 1)) < $ (diff_array - Shift(array, 0, 0, -1)) IF (sb GT 0) THEN diff_array = Smooth(diff_array, sb) ENDIF ELSE BEGIN diff_array = Float(array) diff_array = diff_array < $ (diff_array - Shift(array, 0, 1, 0)) < $ (diff_array - Shift(array, 1, 0, 0)) < $ (diff_array - Shift(array, 0, -1, 0)) < $ (diff_array - Shift(array, -1, 0, 0)) < $ (diff_array - Shift(array, 0, 0, 1)) < $ (diff_array - Shift(array, 0, 0, -1)) IF (sb GT 0) THEN diff_array = Smooth(diff_array, sb) ENDELSE ENDIF ; *** Set up the required variables similar_val = 1B connect_val = 2B c_array = Bytarr(x_size, y_size, z_size) c_array[Where((array GE min_val) AND (array LE max_val))] = similar_val x_size_m1 = x_size - 1L y_size_m1 = y_size - 1L z_size_m1 = z_size - 1L x_ind = xpos y_ind = ypos z_ind = zpos just_found = (z_ind * y_size * x_size) + (y_ind * x_size) + x_ind c_array[just_found] = connect_val num_found = 1L ; *** Start the search IF (diag EQ 0B) THEN BEGIN ; *** No diagonal mode nsew_ind = Lonarr(6, 1) nsew_ind[0, *] = (z_ind * y_size * x_size) + $ (((y_ind + 1L) < y_size_m1) * x_size) + x_ind nsew_ind[1, *] = (z_ind * y_size * x_size) + $ (((y_ind - 1L) > 0L) * x_size) + x_ind nsew_ind[2, *] = (z_ind * y_size * x_size) + $ (y_ind * x_size) + ((x_ind + 1L) < x_size_m1) nsew_ind[3, *] = (z_ind * y_size * x_size) + $ (y_ind * x_size) + ((x_ind - 1L) > 0L) nsew_ind[4, *] = (((z_ind + 1L) < z_size_m1) * y_size * x_size) + $ (y_ind * x_size) + x_ind nsew_ind[5, *] = (((z_ind - 1L) > 0L) * y_size * x_size) + $ (y_ind * x_size) + x_ind cc_array = c_array[nsew_ind[*]] just_found = Where(cc_array EQ similar_val) ; *** Loop while cells are still being found WHILE (just_found[0] GE 0L) DO BEGIN cc_array[just_found] = connect_val c_array[nsew_ind[just_found]] = cc_array[just_found] z_ind = nsew_ind[just_found] / (y_size * x_size) y_ind = (nsew_ind[just_found] - (z_ind * y_size * x_size)) / (x_size) x_ind = (nsew_ind[just_found] - (z_ind * y_size * x_size)) - $ (y_ind * x_size) num_found = N_Elements(just_found) nsew_ind = Lonarr(6, num_found, /Nozero) nsew_ind[0, *] = (z_ind * y_size * x_size) + $ (((y_ind + 1L) < y_size_m1) * x_size) + x_ind nsew_ind[1, *] = (z_ind * y_size * x_size) + $ (((y_ind - 1L) > 0L) * x_size) + x_ind nsew_ind[2, *] = (z_ind * y_size * x_size) + $ (y_ind * x_size) + ((x_ind + 1L) < x_size_m1) nsew_ind[3, *] = (z_ind * y_size * x_size) + $ (y_ind * x_size) + ((x_ind - 1L) > 0L) nsew_ind[4, *] = (((z_ind + 1L) < z_size_m1) * y_size * x_size) + $ (y_ind * x_size) + x_ind nsew_ind[5, *] = (((z_ind - 1L) > 0L) * y_size * x_size) + $ (y_ind * x_size) + x_ind nsew_ind = nsew_ind[Sort(nsew_ind[*])] nsew_ind = nsew_ind[Uniq(nsew_ind)] cc_array = c_array[nsew_ind[*]] IF (range) THEN BEGIN t_array = diff_array[nsew_ind[*]] just_found = Where((cc_array EQ similar_val) AND $ ((t_array GE dec) AND $ (t_array LE inc))) ENDIF ELSE BEGIN just_found = Where(cc_array EQ similar_val) ENDELSE ENDWHILE ENDIF ELSE BEGIN ; *** Diagonal mode nsew_ind = Lonarr(26, 1) nsew_ind[ 0, *] = (z_ind * y_size * x_size) + $ (((y_ind + 1L) < y_size_m1) * x_size) + x_ind nsew_ind[ 1, *] = (z_ind * y_size * x_size) + $ (((y_ind - 1L) > 0L) * x_size) + x_ind nsew_ind[ 2, *] = (z_ind * y_size * x_size) + $ (y_ind * x_size) + ((x_ind + 1L) < x_size_m1) nsew_ind[ 3, *] = (z_ind * y_size * x_size) + $ (y_ind * x_size) + ((x_ind - 1L) > 0L) nsew_ind[ 4, *] = (z_ind * y_size * x_size) + $ (((y_ind + 1L) < y_size_m1) * x_size) + $ ((x_ind + 1L) < x_size_m1) nsew_ind[ 5, *] = (z_ind * y_size * x_size) + $ (((y_ind - 1L) > 0L) * x_size) + $ ((x_ind - 1L) > 0L) nsew_ind[ 6, *] = (z_ind * y_size * x_size) + $ (((y_ind + 1L) < y_size_m1) * x_size) + $ ((x_ind - 1L) > 0L) nsew_ind[ 7, *] = (z_ind * y_size * x_size) + $ (((y_ind - 1L) > 0L) * x_size) + $ ((x_ind + 1L) < x_size_m1) nsew_ind[ 8, *] = (((z_ind + 1L) < z_size_m1) * y_size * x_size) + $ (((y_ind + 1L) < y_size_m1) * x_size) + x_ind nsew_ind[ 9, *] = (((z_ind + 1L) < z_size_m1) * y_size * x_size) + $ (((y_ind - 1L) > 0L) * x_size) + x_ind nsew_ind[10, *] = (((z_ind + 1L) < z_size_m1) * y_size * x_size) + $ (y_ind * x_size) + ((x_ind + 1L) < x_size_m1) nsew_ind[11, *] = (((z_ind + 1L) < z_size_m1) * y_size * x_size) + $ (y_ind * x_size) + ((x_ind - 1L) > 0L) nsew_ind[12, *] = (((z_ind + 1L) < z_size_m1) * y_size * x_size) + $ (((y_ind + 1L) < y_size_m1) * x_size) + $ ((x_ind + 1L) < x_size_m1) nsew_ind[13, *] = (((z_ind + 1L) < z_size_m1) * y_size * x_size) + $ (((y_ind - 1L) > 0L) * x_size) + $ ((x_ind - 1L) > 0L) nsew_ind[14, *] = (((z_ind + 1L) < z_size_m1) * y_size * x_size) + $ (((y_ind + 1L) < y_size_m1) * x_size) + $ ((x_ind - 1L) > 0L) nsew_ind[15, *] = (((z_ind + 1L) < z_size_m1) * y_size * x_size) + $ (((y_ind - 1L) > 0L) * x_size) + $ ((x_ind + 1L) < x_size_m1) nsew_ind[16, *] = (((z_ind - 1L) > 0L) * y_size * x_size) + $ (((y_ind + 1L) < y_size_m1) * x_size) + x_ind nsew_ind[17, *] = (((z_ind - 1L) > 0L) * y_size * x_size) + $ (((y_ind - 1L) > 0L) * x_size) + x_ind nsew_ind[18, *] = (((z_ind - 1L) > 0L) * y_size * x_size) + $ (y_ind * x_size) + ((x_ind + 1L) < x_size_m1) nsew_ind[19, *] = (((z_ind - 1L) > 0L) * y_size * x_size) + $ (y_ind * x_size) + ((x_ind - 1L) > 0L) nsew_ind[20, *] = (((z_ind - 1L) > 0L) * y_size * x_size) + $ (((y_ind + 1L) < y_size_m1) * x_size) + $ ((x_ind + 1L) < x_size_m1) nsew_ind[21, *] = (((z_ind - 1L) > 0L) * y_size * x_size) + $ (((y_ind - 1L) > 0L) * x_size) + $ ((x_ind - 1L) > 0L) nsew_ind[22, *] = (((z_ind - 1L) > 0L) * y_size * x_size) + $ (((y_ind + 1L) < y_size_m1) * x_size) + $ ((x_ind - 1L) > 0L) nsew_ind[23, *] = (((z_ind - 1L) > 0L) * y_size * x_size) + $ (((y_ind - 1L) > 0L) * x_size) + $ ((x_ind + 1L) < x_size_m1) nsew_ind[24, *] = (((z_ind + 1L) < z_size_m1) * y_size * x_size) + $ (y_ind * x_size) + x_ind nsew_ind[25, *] = (((z_ind - 1L) > 0L) * y_size * x_size) + $ (y_ind * x_size) + x_ind cc_array = c_array[nsew_ind[*]] just_found = Where(cc_array EQ similar_val) ; *** Loop while cells are still being found WHILE (just_found[0] GE 0L) DO BEGIN cc_array[just_found] = connect_val c_array[nsew_ind[just_found]] = cc_array[just_found] z_ind = nsew_ind[just_found] / (y_size * x_size) y_ind = (nsew_ind[just_found] - (z_ind * y_size * x_size)) / (x_size) x_ind = (nsew_ind[just_found] - (z_ind * y_size * x_size)) - $ (y_ind * x_size) num_found = N_Elements(just_found) nsew_ind = Lonarr(26, num_found, /Nozero) nsew_ind[ 0, *] = (z_ind * y_size * x_size) + $ (((y_ind + 1L) < y_size_m1) * x_size) + x_ind nsew_ind[ 1, *] = (z_ind * y_size * x_size) + $ (((y_ind - 1L) > 0L) * x_size) + x_ind nsew_ind[ 2, *] = (z_ind * y_size * x_size) + $ (y_ind * x_size) + ((x_ind + 1L) < x_size_m1) nsew_ind[ 3, *] = (z_ind * y_size * x_size) + $ (y_ind * x_size) + ((x_ind - 1L) > 0L) nsew_ind[ 4, *] = (z_ind * y_size * x_size) + $ (((y_ind + 1L) < y_size_m1) * x_size) + $ ((x_ind + 1L) < x_size_m1) nsew_ind[ 5, *] = (z_ind * y_size * x_size) + $ (((y_ind - 1L) > 0L) * x_size) + $ ((x_ind - 1L) > 0L) nsew_ind[ 6, *] = (z_ind * y_size * x_size) + $ (((y_ind + 1L) < y_size_m1) * x_size) + $ ((x_ind - 1L) > 0L) nsew_ind[ 7, *] = (z_ind * y_size * x_size) + $ (((y_ind - 1L) > 0L) * x_size) + $ ((x_ind + 1L) < x_size_m1) nsew_ind[ 8, *] = (((z_ind + 1L) < z_size_m1) * y_size * x_size) + $ (((y_ind + 1L) < y_size_m1) * x_size) + x_ind nsew_ind[ 9, *] = (((z_ind + 1L) < z_size_m1) * y_size * x_size) + $ (((y_ind - 1L) > 0L) * x_size) + x_ind nsew_ind[10, *] = (((z_ind + 1L) < z_size_m1) * y_size * x_size) + $ (y_ind * x_size) + ((x_ind + 1L) < x_size_m1) nsew_ind[11, *] = (((z_ind + 1L) < z_size_m1) * y_size * x_size) + $ (y_ind * x_size) + ((x_ind - 1L) > 0L) nsew_ind[12, *] = (((z_ind + 1L) < z_size_m1) * y_size * x_size) + $ (((y_ind + 1L) < y_size_m1) * x_size) + $ ((x_ind + 1L) < x_size_m1) nsew_ind[13, *] = (((z_ind + 1L) < z_size_m1) * y_size * x_size) + $ (((y_ind - 1L) > 0L) * x_size) + $ ((x_ind - 1L) > 0L) nsew_ind[14, *] = (((z_ind + 1L) < z_size_m1) * y_size * x_size) + $ (((y_ind + 1L) < y_size_m1) * x_size) + $ ((x_ind - 1L) > 0L) nsew_ind[15, *] = (((z_ind + 1L) < z_size_m1) * y_size * x_size) + $ (((y_ind - 1L) > 0L) * x_size) + $ ((x_ind + 1L) < x_size_m1) nsew_ind[16, *] = (((z_ind - 1L) > 0L) * y_size * x_size) + $ (((y_ind + 1L) < y_size_m1) * x_size) + x_ind nsew_ind[17, *] = (((z_ind - 1L) > 0L) * y_size * x_size) + $ (((y_ind - 1L) > 0L) * x_size) + x_ind nsew_ind[18, *] = (((z_ind - 1L) > 0L) * y_size * x_size) + $ (y_ind * x_size) + ((x_ind + 1L) < x_size_m1) nsew_ind[19, *] = (((z_ind - 1L) > 0L) * y_size * x_size) + $ (y_ind * x_size) + ((x_ind - 1L) > 0L) nsew_ind[20, *] = (((z_ind - 1L) > 0L) * y_size * x_size) + $ (((y_ind + 1L) < y_size_m1) * x_size) + $ ((x_ind + 1L) < x_size_m1) nsew_ind[21, *] = (((z_ind - 1L) > 0L) * y_size * x_size) + $ (((y_ind - 1L) > 0L) * x_size) + $ ((x_ind - 1L) > 0L) nsew_ind[22, *] = (((z_ind - 1L) > 0L) * y_size * x_size) + $ (((y_ind + 1L) < y_size_m1) * x_size) + $ ((x_ind - 1L) > 0L) nsew_ind[23, *] = (((z_ind - 1L) > 0L) * y_size * x_size) + $ (((y_ind - 1L) > 0L) * x_size) + $ ((x_ind + 1L) < x_size_m1) nsew_ind[24, *] = (((z_ind + 1L) < z_size_m1) * y_size * x_size) + $ (y_ind * x_size) + x_ind nsew_ind[25, *] = (((z_ind - 1L) > 0L) * y_size * x_size) + $ (y_ind * x_size) + x_ind nsew_ind = nsew_ind[Sort(nsew_ind[*])] nsew_ind = nsew_ind[Uniq(nsew_ind)] cc_array = c_array[nsew_ind[*]] IF (range) THEN BEGIN t_array = diff_array[nsew_ind[*]] just_found = Where((cc_array EQ similar_val) AND $ ((t_array GE dec) AND $ (t_array LE inc))) ENDIF ELSE BEGIN just_found = Where(cc_array EQ similar_val) ENDELSE ENDWHILE ENDELSE ; *** Clean up and return x_ind = 0 y_ind = 0 nsew_ind = 0 cc_array = 0 t_array = 0 diff_array = 0 index = Where(c_array EQ connect_val) RETURN, index END