added additional functions to allow batch reading of ENVI files and conversion o…

…f label images to class images

added additional functions to allow batch reading of ENVI files and conversion o…
…f label images to class images
David Mayerich
1 parent f275c01c
Showing 2 changed files with 48 additions and 7 deletions Show diff stats
python/classify.py
python/envi.py
@@ -16,7 +16,7 @@ import envi
 #generate a 2D color class map using a stack of binary class images
 #input: C is a C x Y x X binary image
 #output: an RGB color image with a unique color for each class
-def classcolor2(C):
+def class2color(C):
  
     #determine the number of classes
     nc = C.shape[0]
@@ -119,6 +119,17 @@ def prob2roc(P, t_vals, mask=[]):
  
     return sklearn.metrics.roc_curve(labels, scores)
  
+#convert a label image to a C x Y x X class image
+def label2class(L):
+    unique = numpy.unique(L)
+    s = L.shape
+    s = numpy.append(numpy.array((len(unique)-1)), s)
+    print(s)
+    C = numpy.zeros(s, dtype=numpy.bool)
+    for i in range(1, len(unique)):
+        C[i-1, :, :] = L == unique[i]
+    return C
+
 #Function to convert a set of class labels to a matrix of neuron responses for an ANN
  
 #Function CNN extraction function
 \ No newline at end of file
@@ -182,12 +182,13 @@ class envi_header:
         f.close()
  
 class envi:
-    def __init__(self, filename, headername = "", maskname = ""):
+    def __init__(self, filename, headername = "", mask = []):
         self.open(filename, headername)
-        if maskname == "":
+        if mask == []:
             self.mask = numpy.ones((self.header.samples, self.header.lines), dtype=numpy.bool)
         else:
-            self.mask = scipy.misc.imread(maskname, flatten=True).astype(numpy.bool)
+            self.mask = mask
+        self.idx = 0                                                               #initialize the batch IDX to 0 for batch reading
  
     def open(self, filename, headername = ""):
         if headername == "":
@@ -232,6 +233,7 @@ class envi:
         M = numpy.zeros((B, P), dtype=self.header.data_type)
         type_bytes = numpy.dtype(self.header.data_type).itemsize
  
+        prev_pos = self.file.tell()
         self.file.seek(0)
         if self.header.interleave == "bip":
             spectrum = numpy.zeros(B, dtype=self.header.data_type)
@@ -262,6 +264,7 @@ class envi:
                 M[:, p:p+i.shape[0]] = plane[:, i]
                 p = p + i.shape[0]
                 bar.update(l+1)
+        self.file.seek(prev_pos)
         return M
  
     def loadband(self, n):
@@ -271,11 +274,12 @@ class envi:
  
         band = numpy.zeros((Y, X), dtype=self.header.data_type)
         type_bytes = numpy.dtype(self.header.data_type).itemsize
-
+        
+        prev_pos = self.file.tell()
         if self.header.interleave == "bsq":
             self.file.seek(n * X * Y * type_bytes)
             self.file.readinto(band)
-
+        self.file.seek(prev_pos)
         return band
  
     #create a set of feature/target pairs for classification
@@ -296,7 +300,33 @@ class envi:
             T.append(t)
  
         return numpy.concatenate(F, 1).transpose(), numpy.concatenate(T)
-
+    
+    #read a batch of data based on the mask
+    def loadbatch(self, npixels):
+        i = numpy.flatnonzero(self.mask)                                      #get the indices of valid pixels
+        npixels = min(npixels, len(i) - self.idx - 1)                        #if there aren't enough pixels, change the batch size
+        B = self.header.bands
+        
+        batch = numpy.zeros((B, npixels), dtype=self.header.data_type)          #allocate space for the batch
+        pixel = numpy.zeros((B), dtype=self.header.data_type)                   #allocate space for a single pixel
+        type_bytes = numpy.dtype(self.header.data_type).itemsize                #calculate the size of a single value
+        if self.header.interleave == "bip":
+            for n in range(0, npixels):                                          #for each pixel in the batch
+                self.file.seek(i[self.idx] * B * type_bytes)                 #seek to the current pixel in the file
+                self.file.readinto(pixel)                                       #read a single pixel
+                batch[:, n] = pixel                                             #save the pixel into the batch matrix
+                self.idx = self.idx + 1
+                #print("idx: " + str(self.idx))
+                #print("tell(): " + str(self.file.tell()))
+            return batch
+        elif self.header.interleave == "bsq":
+            print("ERROR: BSQ batch loading isn't implemented yet!")
+        elif self.header.interleave == "bil":
+            print("ERROR: BIL batch loading isn't implemented yet!")        
+       
+    #returns the current batch index         
+    def getidx(self):
+        return self.idx
  
     def __del__(self):
         self.file.close()
 \ No newline at end of file