Mercurial > repos > mb2013 > nepenthes_3dpca
view ReConstructor.py @ 17:27af4a7b1e1d draft
Uploaded
author | mb2013 |
---|---|
date | Tue, 20 May 2014 03:29:26 -0400 |
parents | fc4bc4ab91b7 |
children |
line wrap: on
line source
# Reconstruction of faces of a .ply file, # based on the symmetry plane # MB import math from math import * import sys import numpy from time import gmtime, strftime # Function main, def main(): var_col = 0 name_file_ply = sys.argv[1] output = open('new_coordinates2.ply', 'w') file_ply = open(name_file_ply) out_log = open(str(sys.argv[15]), 'w') out_log.write('Start time: %s\n\nInput user:\n'%(time())) x_1 = sys.argv[9] x_2 = sys.argv[10] y_1 = sys.argv[11] y_2 = sys.argv[12] z_1 = sys.argv[13] z_2 = sys.argv[14] # to function 'user input side' side = user_input_side(out_log) # to function 'user input sections' number_of_boxes_hor,number_of_boxes_ver,number_of_boxes_z = user_input_sections(out_log) # to function 'user input factor stdev' factor = user_input_factor_stdev(out_log) # to function user input colors col_r, col_g, col_b = user_input_colors(out_log) # to function 'extracting header' var_header, var_vertex_nm,var_face_nm = extracting_header(file_ply) # to funtion 'array coordinates' matrix, matrix2,matrix3 = array_coordinates(name_file_ply, var_vertex_nm, var_header, var_face_nm) # to function 'calc min max' amin,amax = calc_min_max(out_log, matrix, x_1,x_2,y_1,y_2,z_1,z_2) # to function 'range sections' steps_x,steps_y,steps_z,x_window_min_x,x_window_max_x = range_sections(amin,amax, number_of_boxes_hor, number_of_boxes_ver, number_of_boxes_z) # to function 'create list ranges' newlist = create_list_ranges(amin,amax,number_of_boxes_hor, number_of_boxes_ver, number_of_boxes_z, x_window_min_x, x_window_max_x,steps_x,steps_y,steps_z) # to function 'create list coordinates' newlist2, indexlist = create_list_coordinates(newlist) # to function 'fill sections coordinates' newlist, indexlist = fill_sections_coordinates(newlist, newlist2, matrix, indexlist) # to function 'calc sections' difference, differencemean = calc_sections(number_of_boxes_ver,number_of_boxes_hor,number_of_boxes_z,newlist) # to function 'calc mean stdev' mean_percentage, std_percentage = calc_mean_stdev(differencemean,factor) # to function 'calc range' left_range, right_range = calc_range(mean_percentage,std_percentage) # to function 'collect mirror values' total, listindexcount, listindex, listindextotalcount, totalcount = collect_mirror_values(out_log, side, output, right_range, left_range, mean_percentage, std_percentage, newlist, indexlist, difference, number_of_boxes_hor,number_of_boxes_ver,number_of_boxes_z, col_r, col_g, col_b) # to function 'find original faces' m_face,m_face_square = find_original_faces(matrix2,matrix3, listindex) # to function 'replace index values' out_faces = replace_index_values(m_face,listindex,listindextotalcount) out_faces_2 = [] # if there are square faces, do function replace index values again if len(m_face_square)!= 0: out_faces_2 = replace_index_values(m_face_square,listindex,listindextotalcount) output.close() # to function 'write output' write_output(name_file_ply, out_log, totalcount, var_header, var_vertex_nm, var_face_nm,out_faces, out_faces_2) #a = strftime("%a, %d %b %Y %H:%M:%S", gmtime()) out_log.write('End time: %s'%(time())) out_log.close() # close log # Function time def time(): a = strftime("%a, %d %b %Y %H:%M:%S", gmtime()) return a # Function user input side def user_input_side(out_log): # user input side side = int(sys.argv[3]) if side == 0: out_log.write('Left side of object is correct') else: out_log.write('Right side of object is correct') return side # Function user input sections def user_input_sections(out_log): # number of boxes on x axis, this will be multiplied. number_of_boxes_hor = int(sys.argv[4]) # number of boxes on y axis number_of_boxes_ver = int(sys.argv[5]) # number of boxes on z axis number_of_boxes_z = int(sys.argv[6]) out_log.write("number of boxes x axis:\t%s\nnumber of boxes y axis:\t%s\nnumber of boxes z axis:\t%s\n\n"% ((number_of_boxes_hor *2),(number_of_boxes_ver), (number_of_boxes_z))) return number_of_boxes_hor, number_of_boxes_ver, number_of_boxes_z # Function user input colors of reconstructed surface def user_input_colors(out_log): col = sys.argv[8] # four colors can be chosen if int(col) == 0: out_log.write('color is pink\n') col_r = 236 col_g = 149 col_b = 221 elif int(col) == 1: out_log.write('color is blue\n') col_r = 192 col_g = 245 col_b = 250 elif int(col) == 2: out_log.write('color is green\n') col_r = 0 col_g = 255 col_b = 0 else: out_log.write('color is red\n') col_r = 255 col_g = 0 col_b = 0 return col_r, col_g, col_b # Function user input factor standard deviation def user_input_factor_stdev(out_log): factor = int(sys.argv[7]) out_log.write("Factor standard deviation:\t%s\n\n"%(factor)) return factor # Function extracting values of header ply file def extracting_header(file_ply): for x in range(0, 20): readheader = file_ply.readline().strip().split() if readheader[0] == 'end_header': # when the words 'end_header' are found var_header = x # var_header is line number if readheader[0] == "element" and readheader[1] == "vertex": # when 'element vertex' found var_vertex_nm = readheader[2] # amount of vertexen if readheader[0] == "element" and readheader[1] == "face": # when 'element face' found var_face_nm = readheader[2] # amount of faces file_ply.close() return var_header, var_vertex_nm,var_face_nm # Function extracting coordintates of ply file def array_coordinates(name_file_ply, var_vertex_nm, var_header,var_face_nm): file1 = open(name_file_ply) matrix = numpy.zeros((int(var_vertex_nm),3)) # creating empty numpy array of amount of vertexen matrix2 = numpy.empty((int(var_face_nm),3)) # creating empty numpy array of amount of triangle faces matrix3 = numpy.empty((int(var_face_nm),4)) # creating empty numpy array of amounf of square faces 4 count = 0 # counter for vertexen count2 = 0 # counter for faces # Putting all the vertexen and faces in a matrix for a in range(0, (int(var_header) + int(var_vertex_nm) + int(var_face_nm) + 1)): line = file1.readline().strip().split() # reading every line if int(var_header) < a < (int(var_vertex_nm)+ int(var_header)+1): #vertexen matrix[count][0] = float(line[0]) # x coordinate matrix[count][1] = float(line[1]) # y coordinate matrix[count][2] = float(line[2]) # z coordinate count += 1 # counter + 1 for next line in matrix if a > (int(var_vertex_nm)+ int(var_header)): # faces if len(line) == 4: # if triangle faces exists matrix2[count2][0] = int(line[1]) # first coordinate for face matrix2[count2][1] = int(line[2]) # second coordinate for face matrix2[count2][2] = int(line[3]) # third coordinate for face #check1 = True if len(line) == 5: # if square faces exists matrix3[count2][0] = int(line[1]) # first coordinate for face matrix3[count2][1] = int(line[2]) # second coordinate for face matrix3[count2][2] = int(line[3]) # third coordinate for face matrix3[count2][3] = int(line[4]) # fourth coordinate for face #check2 = True count2 += 1 # counter + 1 for next line in matrix file1.close() # to function calc_histogram calc_histogram(matrix) return matrix,matrix2,matrix3 # Function histogram def calc_histogram(matrix): # histogram of x,y, z coordinates out_histo = open(str(sys.argv[16]), 'w') # open output x_co = numpy.array(matrix[:,0]) histo_x = numpy.histogram(x_co, bins = 50) # calculate x coordinate distribution y_co = numpy.array(matrix[:,1]) histo_y = numpy.histogram(y_co, bins = 50) # calculate y coordinate distribution z_co = numpy.array(matrix[:,2]) histo_z = numpy.histogram(z_co, bins = 50) # calculate z coordinate distribution # write histogram output out_histo.write("x nm point\tx coordinate\ty nm point\ty coordinate\tz nm point\tz coordinate\n") for item in range(0,len(histo_x[0])): out_histo.write("%s\t%s\t\t%s\t%s\t\t%s\t%s\n"%(histo_x[0][item], histo_x[1][item], histo_y[0][item], histo_y[1][item], histo_z[0][item], histo_z[1][item], )) out_histo.close() #close histogram output # Function calculation minimum and maximum of x,y and z def calc_min_max(out_log, matrix, x_1,x_2,y_1,y_2,z_1,z_2): amin = numpy.amin(matrix, axis = 0) #minima of x y and z amax = numpy.amax(matrix, axis = 0) #maxima of x y and z # if user input max and min values, change the min and max if x_1 != 'standard': amin[0] = int(x_1) if x_2 != 'standard': amax[0] = int(x_2) if y_1 != 'standard': amin[1] = int(y_1) if y_2 != 'standard': amax[1] = int(y_2) if z_1 != 'standard': amin[2] = int(z_1) if z_2 != 'standard': amax[2] = int(z_2) # write to log out_log.write("lowest x:\t%s\nhighest x:\t%s\nlowest y:\t%s\nhighest y:\t%s\nlowest z:\t%s\nhighest z:\t%s\n\n" %(amin[0],amax[0],amin[1],amax[1],amin[2],amax[2])) return amin, amax # Function calculating range for sections def range_sections(amin,amax,number_of_boxes_hor, number_of_boxes_ver,number_of_boxes_z): # Find the highest absolute x because the boxes on the x axis has to be mirrored in the symmetry plane if abs(amax[0]) < abs(amin[0]): # if the highest x is absolute smaller than the lowest x: x_window_min_x = amin[0] # lowest x is lowest windows x x_window_max_x= amin[0] *-1 # lowest x * -1 is highest windows x steps_x = abs(amin[0]) / number_of_boxes_hor # length of x boxes else: x_window_min_x = amax[0] *-1 # highest x *-1 is lowest windows x x_window_max_x = amin[0] # highest x is highest window x steps_x = abs(amax[0]) / number_of_boxes_hor # length of x boxes #Calculating range of y boxes steps_y = (abs(amax[1]) + abs(amin[1])) / number_of_boxes_ver #Calculating range of z boxes steps_z = (abs(amax[2]) + abs(amin[2])) / number_of_boxes_z return steps_x, steps_y, steps_z, x_window_min_x,x_window_max_x # Function create list ranges of each section def create_list_ranges(amin,amax,number_of_boxes_hor, number_of_boxes_ver,number_of_boxes_z, x_window_min_x, x_window_max_x,steps_x,steps_y,steps_z): # creat list with ranges of each section #creating the list with the ranges for every section y_window_min = amin[1] y_window_max = amin[1] z_window_min = amin[2] newlist = [] sublist = [] y_window_min2 = y_window_min z_window_min2 = z_window_min # From front to back, from floor to ceiling for a in range(0,int(number_of_boxes_ver)): #number of rows x_window_min = x_window_min_x x_window_min2 = x_window_min_x + steps_x y_window_min = y_window_min2 y_window_min2+= steps_y z_window_min = amin[2] z_window_min2 = z_window_min for c in range(0,int(number_of_boxes_z)): # in z direction z_window_min = z_window_min2 z_window_min2 += steps_z x_window_min = x_window_min_x x_window_min2 = x_window_min_x + steps_x for b in range(0, int((2*number_of_boxes_hor))): #number of columns sublist.append(x_window_min) sublist.append(x_window_min2) sublist.append(y_window_min) sublist.append(y_window_min2) sublist.append(z_window_min) sublist.append(z_window_min2) newlist.append(sublist) sublist = [] x_window_min = x_window_min2 x_window_min2 += steps_x return newlist # Function create list coordinates in sections def create_list_coordinates(newlist): # create list coordinates in sections #empty the list but maintaining the structure newlist2=[] newlist2sub = [] newlistsub = [] for x in range(0,len(newlist)): for y in range(5,-1,-1): newlist2sub.append(newlist[x].pop(y)) newlist2.append(newlist2sub) newlist2sub = [] # creating empty index list for storing line index of coordinates indexlist = [] for x in range(0,len(newlist)): indexlist.append([]) return newlist2, indexlist # Function fill list sections with coordinates def fill_sections_coordinates(newlist,newlist2,matrix,indexlist): # fill list sections with coordinates #Filling the sections with the coordinates for y in range(0,len(newlist2)): #notice the ranges are in reverse order of each section rows_vertex = (numpy.where((matrix[:,0]>= float(newlist2[y][5])) & (matrix[:,0] < float(newlist2[y][4])) & (matrix[:,1] >= float(newlist2[y][3])) & (matrix[:,1] < float(newlist2[y][2])) & (matrix[:,2] >= float(newlist2[y][1])) & (matrix[:,2] < float(newlist2[y][0])))) vertex_new = matrix[list(rows_vertex)] newlist[y].append((vertex_new)) indexlist[y].append((rows_vertex)) return newlist, indexlist # Function calculate mean and number of coordinates sections def calc_sections(number_of_boxes_ver,number_of_boxes_hor,number_of_boxes_z,newlist): #Selecting the sections on the symmetry axis and calculating the mean of every section counter = 0 counter2 = 0 counter3 = 0 row_count = 0 difference = [] #whit sections which are empty differencemean = [] #without sections which are empty total = 0 for x in range(0, (int(number_of_boxes_ver)*int(number_of_boxes_hor) * int(number_of_boxes_z))): # through all the lists/sections if counter % (int (number_of_boxes_hor)*2) == 0: counter = 0 if counter3 == (int(number_of_boxes_hor)): counter2 += (int(number_of_boxes_hor)) counter3 = 0 first_pos = (newlist[counter2][0]) # coordinates of left section last_pos = (newlist[counter2 + (int(number_of_boxes_hor)*2 -1)- counter][0]) # coordinates of right section if len(first_pos) == 0: # If there are no coordinates in a section number_coordinates_box1 = len(first_pos) mean1 = 2000 mean3 = '' else: number_coordinates_box1 = len(first_pos) for e in range(0,len(first_pos)): # If there are coordinates in a section total += float(first_pos[e][0]) mean1 = total/len(first_pos) mean3 = mean1 total = 0 if len(last_pos) == 0: # If there are no coordinates in a section number_coordinates_box2 = len(last_pos) mean2 = 1000 mean4 = '' else: number_coordinates_box2 = len(last_pos) for e in range(0,len(last_pos)): # If there are coordinates in a section total += float(last_pos[e][0]) mean2 = total/len(last_pos) mean4 = mean2 total = 0 number_coordinates_box1 = len(first_pos) number_coordinates_box2 = len(last_pos) # if number of coordinates in sections deviate to much, if (abs(number_coordinates_box1 - number_coordinates_box2) > (number_coordinates_box1 / 2.0) ) and(abs(number_coordinates_box1 - number_coordinates_box2) > (number_coordinates_box2 / 2.0)): mean1 = 3000 mean2 = 4000 difference.append(float(abs(mean2 + mean1))) #the means of every sections counter += 2 counter2 += 1 counter3 += 1 try: differencemean.append(abs(mean4 + mean3)) # the mean without empty sections except: continue return difference, differencemean # Function calculating the mean and the standard deviation def calc_mean_stdev(differencemean, factor): mean_percentage = numpy.mean(differencemean) #calculating the mean without empty sections std_percentage = numpy.std(differencemean) # calculating the standard deviation of the list without the empty sections std_percentage = std_percentage * factor return mean_percentage, std_percentage # Function calculating range of accepted differences def calc_range(mean_percentage,std_percentage): left_range = float(mean_percentage) - float(std_percentage) #left range mean minus one standard deviation right_range = float(mean_percentage) + (float(std_percentage) )# right range mean plus one standard deviation if left_range > 0: left_range = 0 return left_range, right_range # Function collecting incorrect coordinates def collect_mirror_values(out_log, side, output, right_range, left_range,mean_percentage, std_percentage, newlist, indexlist,difference, number_of_boxes_hor,number_of_boxes_ver,number_of_boxes_z, col_r, col_g, col_b): # Collecting the values of the sections which have to be mirrored out_log.write('mean x difference:\t%s\nstandarddeviation mean x difference:\t%s\nrange:\t%s - %s\n\n'% ((mean_percentage),(std_percentage),(left_range),(right_range))) y = 0 counter = 0 counter2 = 0 sub = [] total = [] left_range = "%.10f"%(left_range) indexcount = 0 facecount = 0 totalcount = 0 listindex = [] listindexsub = [] listindexcount = [] listindextotalcount = [] for x in range(0, len(difference)): difference_x = "%.10f"%(difference[x]) if counter %(int(number_of_boxes_hor)) == 0 and x != 0: counter = 0 counter2 += (int(number_of_boxes_hor)) counter += 1 counter2 += 1 #left side counter # if left side of the object is correct if side == 0: if (float(difference_x) < float(left_range)) or (float(difference_x) > float(right_range)): if len(newlist[counter2 -1][0]) != 0: sub = (newlist[counter2 -1][0]) #coordinates in list for y in range(0, len(sub)): i = sub[y] index_a = indexlist[counter2 -1][0][0][y] try: listindex.append(int(index_a)) # index number listindextotalcount.append(int(totalcount)) listindexsub.append(int(index_a)) # index number listindexsub.append(totalcount) # index number of new coordinate except: continue listindexcount.append(listindexsub) # index number original listindexsub = [] number= float(sub[y][0]) * -1 # x coordinate * -1 for mirroring at symmetry plane sub[y][0] = number total.append(sub[y]) totalcount += 1 # writing new coordinates to output output.write('%s %s %s %s %s %s\n'%(sub[y][0], float(sub[y][1]), float(sub[y][2]),(col_r), (col_g), (col_b))) # if right side of the object is correct if side == 1: dif = (int(number_of_boxes_hor)) - counter counter3 = counter2 + (dif*2) + 1 # right side counter if (float(difference[x] )> float(right_range)) or (float(difference[x]) < float(left_range)): if len(newlist[counter3 -1][0]) != 0: sub = (newlist[counter3 -1][0]) # coordinates in list for y in range(0, len(sub)): i = sub[y] index_a = indexlist[counter3 -1][0][0][y] try: listindex.append(int(index_a))# index number listindextotalcount.append(int(totalcount)) listindexsub.append(int(index_a)) # index number listindexsub.append(totalcount)# index number of new coordinate except: continue listindexcount.append(listindexsub)# index number original listindexsub = [] number= float(sub[y][0]) * -1 # x coordinate * -1 for mirroring at symmetry plane sub[y][0] = number total.append(sub[y]) totalcount += 1 # writing new coordinates to output output.write('%s %s %s %s %s %s\n'%(sub[y][0], float(sub[y][1]), float(sub[y][2]),(col_r), (col_g), (col_b))) return total, listindexcount, listindex, listindextotalcount,totalcount # Function extract original faces for corrected coordinates def find_original_faces(matrix2, matrix3, listindex): # finding the original faces of kopied points ix = numpy.in1d(matrix2.ravel(),listindex).reshape(matrix2.shape) # finding where the listindex is the same as the matrix rows, cols = numpy.where(ix) # finding index of the faces m_face = matrix2[list(set(rows))] # extracting the rows with those faces m_face_square = [] # if square faces present if matrix3[-1][0] != 0: ix_square = numpy.in1d(matrix3.ravel(),listindex).reshape(matrix3.shape) # finding where the listindex is the same as the matrix rows_square, cols3 = numpy.where(ix_square) # finding index of the faces m_face_square = matrix3[list(set(rows_square))] # extracting the rows with those faces return m_face, m_face_square # Function replacing index values def replace_index_values(m_face,listindex,listindextotalcount): c = [] array_oi = numpy.array(m_face) maxn = numpy.amax(array_oi) palette = list(range(int(maxn))) palette = numpy.array(palette) key = numpy.array(listindex) listindextotalcount = numpy.array(listindextotalcount) # sorting the lists which have to be replaced order= numpy.argsort(key) # sorting of the array key litc_sorted = listindextotalcount[order] # sorting the palette key_sorted = key[order] # sorting the key palette2 = list(range(int(maxn))) # creating list with all possible values in it with max the highest number in the faces key2 = [] counter = 0 for item in range(0,int(maxn)): # make array containing what have to be changed if (counter < len(key_sorted)) and item == (key_sorted[counter]): key2.append(litc_sorted[counter]) counter += 1 else: key2.append(numpy.nan) key2 = numpy.array(key2) # converten of list to array index_f = numpy.digitize(array_oi.reshape(-1,), palette2)-1 # indexing which numbers has to be changed out_sub =(key2[index_f].reshape(array_oi.shape)) # new array creating with changed values out_faces = out_sub[~numpy.isnan(out_sub).any(axis=1)] # extracting the faces with not changed values, (nan) return out_faces # Function write output def write_output(name_file_ply, out_log, totalcount, var_header, var_vertex_nm, var_face_nm,out_faces,out_faces_2): # write output outfile2 = open(str(sys.argv[2]), 'w')#writing the output file file1 = open(name_file_ply) g = 0 for d in range(0,(int(var_header) + int(var_vertex_nm) + int(var_face_nm) + 1)): line2 = file1.readline().strip() readline2 = line2.strip().split() if readline2[0] == "element" and readline2[1] == "vertex": outfile2.write("element vertex %s\n"%(int(var_vertex_nm) + totalcount)) #number of vertexen changing elif readline2[0] == "element" and readline2[1] == "face": outfile2.write("element face %s\n"%(int(var_face_nm) + len(out_faces) + len(out_faces_2)))#number of faces changing elif (int(var_header) < d < (int(var_header) + int(var_vertex_nm))): #rotated vertexen, with original color code outfile2.write('%s\n'%(line2)) g += 1 elif (int(var_header) + int(var_vertex_nm)) == d: #for new vertexen, with color code 0,0,0 outfile2.write('%s\n'%(line2)) #the last original vertex g += 1 with open('new_coordinates2.ply') as infile: for line in infile: outfile2.write(line) else: #everything left outfile2.write('%s\n'%(line2)) #writing new faces to output for z in range(0,len(out_faces)): outfile2.write("3 %s %s %s\n"%((int(out_faces[z][0])+ int(var_vertex_nm)), (int(out_faces[z][1])+ int(var_vertex_nm)), (int(out_faces[z][2])+ int(var_vertex_nm)))) #writing new square faces to output if len(out_faces_2) != 0: for x in range(0,len(out_faces_2)): outfile2.write("4 %s %s %s %s\n"%((int(out_faces_2[z][0])+ int(var_vertex_nm)), (int(out_faces_2[z][1])+ int(var_vertex_nm)), (int(out_faces_2[z][2])+ int(var_vertex_nm)), (int(out_faces_2[z][3] + int(var_vertex_nm))))) out_log.write('number of reconstructed faces:\t%s\n\n'%(len(out_faces) + len(out_faces_2))) outfile2.close() main()