Source Code for Module pyvision.types.RangeImage

  1  # PyVision License 
  2  # 
  3  # Copyright (c) 2009 David S. Bolme 
  4  # All rights reserved. 
  5  # 
  6  # Redistribution and use in source and binary forms, with or without 
  7  # modification, are permitted provided that the following conditions 
  8  # are met: 
  9  #  
 10  # 1. Redistributions of source code must retain the above copyright 
 11  # notice, this list of conditions and the following disclaimer. 
 12  #  
 13  # 2. Redistributions in binary form must reproduce the above copyright 
 14  # notice, this list of conditions and the following disclaimer in the 
 15  # documentation and/or other materials provided with the distribution. 
 16  #  
 17  # 3. Neither name of copyright holders nor the names of its contributors 
 18  # may be used to endorse or promote products derived from this software 
 19  # without specific prior written permission. 
 20  #  
 21  #  
 22  # THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDERS AND CONTRIBUTORS 
 23  # ``AS IS'' AND ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT 
 24  # LIMITED TO, THE IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR 
 25  # A PARTICULAR PURPOSE ARE DISCLAIMED.  IN NO EVENT SHALL THE REGENTS OR 
 26  # CONTRIBUTORS BE LIABLE FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, 
 27  # EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT LIMITED TO, 
 28  # PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE, DATA, OR 
 29  # PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY THEORY OF 
 30  # LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT (INCLUDING 
 31  # NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE OF THIS 
 32  # SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE. 
 33   
 34   
 35  import gzip 
 36  import numpy as np 
 37  import time 
 38  import pyvision as pv 
 39  import scipy.interpolate as it 
 40  import scipy.ndimage as nd 
 41   
 42 -class RangeImage: 
 43      ''' 
 44      This class is used to handle range images. Originally written to handle 
 45      output from the Minolta Vivid sensors distributed with the Face Recognition 
 46      Grand Challenge 2004 
 47       
 48      This implementation currently can parse range images in ".abs" or ".abs.gz" format. 
 49       
 50      Very little type checking is done during parsing so unexpected exception or 
 51      unusual behavior may occur if the file is not formated properly. 
 52       
 53      This is a sample for the .abs file format: 
 54      480 rows 
 55      640 columns 
 56      pixels (flag X Y Z): 
 57      0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0  ... 
 58      -999999.000000 -999999.000000 -999999.000000 -9999 ... 
 59      -999999.000000 -999999.000000 -999999.000000 -9999 ... 
 60      -999999.000000 -999999.000000 -999999.000000 -9999 ... 
 61       
 62      Author: David S. Bolme 2009 
 63      ''' 
 64       
 65 -    def __init__(self,filename): 
 66          ''' 
 67          Reads a file containing range data. 
 68          ''' 
 69   
 70          if filename[-7:] == '.abs.gz': 
 71              # Assume the date is a zlib compressed abs file  
 72              f = gzip.open(filename) 
 73               
 74          if filename[-4:] == '.abs': 
 75              # Assume uncompressed 
 76              f = open(filename) 
 77           
 78          #print buffer[:100] 
 79           
 80          #lines = buffer.split(EOL) 
 81          rows = int(f.next().split()[0]) 
 82          cols = int(f.next().split()[0]) 
 83          #format = f.next() 
 84           
 85          self.width = cols 
 86          self.height = rows 
 87           
 88          self.flags = np.array([int(v) for v in f.next().split()]).reshape(rows,cols).transpose() 
 89          self.x = np.array([float(v) for v in f.next().split()]).reshape(rows,cols).transpose() 
 90          self.y = np.array([float(v) for v in f.next().split()]).reshape(rows,cols).transpose() 
 91          self.z = np.array([float(v) for v in f.next().split()]).reshape(rows,cols).transpose() 
 92           
 93           
 94 -    def getRange(self): 
 95          ''' 
 96          @returns: xmin,xmax,ymin,ymax,zmin,zmax 
 97          ''' 
 98           
 99          flags = np.array(self.flags.flatten(),dtype=np.bool) 
100   
101          X = self.x.flatten()[flags] 
102          Y = self.y.flatten()[flags] 
103          Z = self.z.flatten()[flags] 
104          return min(X),max(X),min(Y),max(Y),min(Z),max(Z)     
105       
106       
107 -    def getXImage(self): 
108          ''' 
109          @returns: the x coordinates. 
110          ''' 
111          xmin,_,_,_,_,_ = self.getRange() 
112           
113          r,c = self.x.shape  
114           
115          flags = np.array(self.flags.flatten(),dtype=np.bool) 
116          X = self.x.flatten().copy() 
117          X[ flags != True ] = xmin 
118           
119          X = X.reshape(r,c) 
120           
121          return pv.Image(X) 
122           
123           
124 -    def getYImage(self): 
125          ''' 
126          @returns: the y coordinates. 
127          ''' 
128          _,_,ymin,_,_,_ = self.getRange() 
129           
130          r,c = self.x.shape  
131           
132          flags = np.array(self.flags.flatten(),dtype=np.bool) 
133          Y = self.y.flatten().copy() 
134          Y[ flags != True ] = ymin 
135           
136          Y = Y.reshape(r,c) 
137           
138          return pv.Image(Y) 
139   
140   
141 -    def getZImage(self): 
142          ''' 
143          @returns: the z coordinates. 
144          ''' 
145          _,_,_,_,zmin,_ = self.getRange() 
146           
147          r,c = self.x.shape  
148           
149          flags = np.array(self.flags.flatten(),dtype=np.bool) 
150          Z = self.z.flatten().copy() 
151          Z[ flags != True ] = zmin 
152           
153          Z = Z.reshape(r,c) 
154           
155          return pv.Image(Z) 
156   
157 -    def getMaskImage(self): 
158          ''' 
159          @returns: the missing value mask. 
160          ''' 
161          _,_,_,_,zmin,_ = self.getRange() 
162           
163          r,c = self.x.shape  
164           
165          flags = np.array(self.flags.flatten(),dtype=np.bool) 
166          Z = self.z.flatten().copy() 
167          Z[ flags != True ] = zmin 
168           
169          Z = Z.reshape(r,c) 
170           
171          return pv.Image(Z) 
172       
173 -    def populateMissingData(self,approach="Smooth",ilog=None): 
174          ''' 
175          This function is used to interpolate missing data in the image. 
176          ''' 
177          if approach == 'Smooth': 
178              # first run a median filter over the array, then smooth the result. 
179              #xmin,xmax,ymin,ymax,zmin,zmax = self.getRange() 
180              mask = np.array(self.flags,dtype=np.bool) 
181               
182              z = self.getZImage().asMatrix2D() 
183              median = nd.median_filter(z,size=(15,15)) 
184   
185              mask = mask.flatten() 
186              z = z.flatten() 
187              median = median.flatten() 
188   
189              z[ mask==False ] = median[ mask==False ] 
190               
191              if ilog != None: 
192                  ilog.log(pv.Image(median.reshape(self.width,self.height)),label="Median") 
193                  ilog.log(pv.Image(z.reshape(self.width,self.height)),label="ZMedian") 
194               
195              mask = mask.flatten() 
196              z = z.flatten() 
197              median = median.flatten() 
198               
199              for i in range(5): 
200                  tmp = z.copy() 
201                  smooth = nd.gaussian_filter(z.reshape(self.width,self.height),2.0).flatten() 
202                  z[ mask==False ] = smooth[ mask==False ] 
203                  print "Iteration:",i,(z-tmp).max(),(z-tmp).min() 
204                  ilog.log(pv.Image(z.reshape(self.width,self.height)),label="ZSmooth%02d"%i) 
205                  ilog.log(pv.Image((z-tmp).reshape(self.width,self.height)),label="ZSmooth%02d"%i) 
206                   
207                   
208          if approach == 'RBF': 
209              mask = np.array(self.flags,dtype=np.bool) 
210              mask = mask.flatten() 
211       
212              x = np.arange(self.width).reshape(self.width,1) 
213              x = x*np.ones((1,self.height)) 
214              x = x.flatten() 
215       
216              y = np.arange(self.height).reshape(1,self.height) 
217              y = y*np.ones((self.width,1)) 
218              y = y.flatten() 
219               
220              z = self.z.copy() 
221              z = z.flatten() 
222               
223              print "Coords:" 
224              print len(mask) 
225              print len(x[mask]) 
226              print len(y[mask]) 
227              print len(z[mask]) 
228               
229              # this produces an error.  Probably has too much data 
230              it.Rbf(x[mask],y[mask],z[mask]) 
231              pass 
232           
233           
234                   
235               
236               
237           
238  if __name__ == "__main__": 
239      ilog = pv.ImageLog() 
240      filename = "02463d562.abs.gz" 
241      im = pv.Image("02463d563.ppm") 
242       
243      t = time.time() 
244      ri = RangeImage(filename) 
245      t = time.time() - t 
246      print t 
247       
248      print ri.getRange() 
249      ilog.log(ri.getXImage(),"X_Image") 
250      ilog.log(ri.getYImage(),"Y_Image") 
251      ilog.log(ri.getZImage(),"Z_Image") 
252      ilog.log(im,"Color") 
253       
254      ri.populateMissingData(ilog=ilog) 
255       
256      ilog.show() 
257