Normalisation and core finding improved

Author: MikeHughesKent

docs/source/pybundle_class.rst

@@ -27,15 +27,15 @@ for a detailed description of each option's meaning.
 * coreMethod = None (``set_core_method``)
 * outputType = 'float64' (``set_output_type``)
 
-**CROP/MASK Settings (for FILTER/EDGE_FILTER only):**
+**CROP/MASK Settings:**
 
 * applyMask = False (``set_apply_mask``)
 * autoMask = True (``set_auto_mask``)
 * autoLoc = False (``set_auto_loc``)
 * crop = False(``set_crop``)
 * loc = None (``set_loc``)
 * mask = None (``set_mask``)
-
+* radius = None (``set_radius``)
 
 **CALIB/BACKGROUND/NORMALISATION Settings:**
 

pyproject.toml

@@ -6,7 +6,7 @@ build-backend = "setuptools.build_meta"
 
 [project]
 name = "PyFibreBundle"
-version = "1.3.1"
+version = "1.3.2"
 description = "Image processing of images acquired through fibre imaging bundle, including core removal, mosaicing and super-resolution.."
 readme = "README.md"
 authors = [{ name = "Michael Hughes", email = "m.r.hughes@kent.ac.uk" }]

src/pybundle/core.py

@@ -260,7 +260,9 @@ def find_bundle(img, **kwargs):
 
 
 def crop_rect(img, loc):
-    """Extracts a square around the bundle using specified co-ordinates.
+    """Extracts a square around the bundle using specified co-ordinates. If the
+    rectange is larger than the image then the returned image will be a rectangle,
+    limited by the extent of the image.
     
     Returns tuple of (cropped image as 2D numpy array, new location tuple)
 
@@ -269,16 +271,27 @@ def crop_rect(img, loc):
         loc  : location to crop, specified as bundle location tuple of 
                (centreX, centreY, radius)
     """
-    
-    cx,cy, rad = loc
-    imgCrop = img[cy-rad:cy+ rad, cx-rad:cx+rad]
-    
-    # Correct the co-ordinates of the bundle so that they
-    # are correct for new cropped image
-    newLoc = [rad,rad,loc[2]]
-   
-    return imgCrop, newLoc
-
+    if loc is not None:
+        
+        h,w = np.shape(img)[:2]
+        cx,cy, rad = loc
+        
+        minX = np.clip(cx-rad, 0, None)
+        maxX = np.clip(cx+rad, None, w)
+        
+         
+        minY = np.clip(cy-rad, 0, None)
+        maxY = np.clip(cy+rad, None, h)
+        
+        imgCrop = img[minY:maxY, minX: maxX]
+        
+        # Correct the co-ordinates of the bundle so that they
+        # are correct for new cropped image
+        newLoc = [rad,rad,loc[2]]
+       
+        return imgCrop, newLoc
+    else:
+        return img, None
 
 
 
@@ -314,14 +327,17 @@ def apply_mask(img, mask):
                  with areas to be kept as 1 and areas to be masked as 0.
     """
 
-    if img.ndim == 3:
-        m = np.expand_dims(mask, 2)
+    if mask is not None:
+        if img.ndim == 3:
+            m = np.expand_dims(mask, 2)
+        else:
+            m = mask
+        imgMasked = np.multiply(img, m)
+        
+        return imgMasked
     else:
-        m = mask
-    imgMasked = np.multiply(img, m)
+        return img
 
-    return imgMasked
-  
 
 def auto_mask(img, loc = None, **kwargs):
     """ Locates bundle and sets pixels outside to 0 .
@@ -332,13 +348,24 @@ def auto_mask(img, loc = None, **kwargs):
     Keyword Arguments:
         loc    : optional location of bundle as tuple of (centreX, centreY, radius), 
                  defaults to determining this using find_bundle
+        radius : optional, int, radius of mask to use rather than the automatically
+                 determined radius        
         Others : if loc is not specified, other optional keyword arguments will
                  be passed to find_bundle.
 
 
     """
+    radius = kwargs.get('radius', None)
+    
+    # If location not specified, find it
     if loc is None:
         loc = pybundle.find_bundle(img, **kwargs)
+    
+    # If radius was specified, replace auto determined radius
+    if radius is not None:
+        loc = (loc[0], loc[1], radius)
+        
+    # Mask image    
     mask = pybundle.get_mask(img, loc)
     imgMasked = pybundle.apply_mask(img, mask)
 

src/pybundle/core_interpolation.py

@@ -14,6 +14,9 @@
 import math
 import time
 
+
+import matplotlib.pyplot as plt
+
 import cv2 as cv
 
 from scipy.spatial import Delaunay
@@ -53,15 +56,18 @@ def find_cores(img, coreSpacing):
      # If a colour image, convert to greyscale by taking the maximum value across the channels
      imgF = pybundle.max_channels(imgF)
 
-     imgF = (imgF / np.max(imgF) * 255).astype('uint8')
+     imgF = (imgF / np.max(imgF) * 255).astype('uint8') 
 
      # Find regional maximum by taking difference between dilated and original
      # image. Because of the way dilation works, the local maxima are not changed
      # and so these will have a value of 0
-     kernel = cv.getStructuringElement(cv.MORPH_ELLIPSE, (coreSpacing,coreSpacing))
+     kernel = cv.getStructuringElement(cv.MORPH_ELLIPSE, (int(round(coreSpacing)),int(round(coreSpacing)) ))
+     kernel = cv.getStructuringElement(cv.MORPH_ELLIPSE, (3,3 ))
+
      imgD = cv.dilate(imgF, kernel)
+     
      imgMax = 255 - (imgF - imgD)  # we need to invert the image
-
+     
      # Just keep the maxima
      thres, imgBinary = cv.threshold(imgMax, 0,1,cv.THRESH_BINARY+cv.THRESH_OTSU)
 
@@ -117,9 +123,10 @@ def core_values(img, coreX, coreY, filterSize, **kwargs):
 
     if filterSize is not None:
         img = pybundle.g_filter(img, filterSize)
-        
-    if numba and numbaAvailable:
-        cInt = core_value_extract_numba(img, coreX, coreY)
+
+    cInt = np.zeros(np.shape(coreX))
+    if numba and numbaAvailable:     
+        cInt = cInt + core_value_extract_numba(img, coreX, coreY)         
     else:
         cInt = img[coreY, coreX]
 
@@ -156,29 +163,31 @@ def calib_tri_interp(img, coreSize, gridSize, **kwargs):
                      spurious core detections outside of bundle, defualts to True
         mask       : optional, boolean, when reconstructing output image will be masked outside of 
                      bundle, defaults to True
+        
     """
 
-    centreX = kwargs.get('centreX', -1)
-    centreY = kwargs.get('centreY', -1)
-    radius = kwargs.get('radius', -1)
+    centreX = kwargs.get('centreX', None)
+    centreY = kwargs.get('centreY', None)
+    radius = kwargs.get('radius', None)
     filterSize = kwargs.get('filterSize', 0)
     normalise = kwargs.get('normalise', None)
     autoMask = kwargs.get('autoMask', True)
     mask = kwargs.get('mask', True)
     background = kwargs.get('background', None)
+    
     if autoMask:
-        img = pybundle.auto_mask(img)
+        img = pybundle.auto_mask(img, radius = radius)
     # Find the cores in the calibration image
     coreX, coreY = pybundle.find_cores(img, coreSize)
     coreX = np.round(coreX).astype('uint16')
     coreY = np.round(coreY).astype('uint16')
 
     # Default values
-    if centreX < 0:
+    if centreX is None:
         centreX = np.mean(coreX)
-    if centreY < 0:
+    if centreY is None:
         centreY = np.mean(coreY)
-    if radius < 0:
+    if radius is None:
         dist = np.sqrt((coreX - centreX)**2 + (coreY - centreY)**2)
         radius = max(dist)
 
@@ -385,9 +394,10 @@ def recon_tri_interp(img, calib, **kwargs):
      numba = kwargs.get('numba', True)
 
      # Extract intensity from each core
+     t1 = time.perf_counter()
      cVals = pybundle.core_values(
          img, calib.coreX, calib.coreY, calib.filterSize, **kwargs).astype('float64')
-
+     
      if calib.background is not None:
          cVals = cVals - calib.backgroundVals