Loading autocnet/matcher/naive_template.py +2 −2 Original line number Diff line number Diff line Loading @@ -130,7 +130,7 @@ def pattern_match(template, image, upsampling=8, metric=cv2.TM_CCOEFF_NORMED, er # Pad the result array with values outside the valid correlation range width = (np.asarray(u_template.shape) - np.asarray(corrmap.shape)) // 2 print(width) result = np.pad(corrmap, width, mode='constant') # pads zeros if metric == cv2.TM_SQDIFF or metric == cv2.TM_SQDIFF_NORMED: matched_y, matched_x = np.unravel_index(np.argmin(result), result.shape) Loading autocnet/matcher/subpixel.py +20 −6 Original line number Diff line number Diff line Loading @@ -270,11 +270,23 @@ def subpixel_template(reference_roi, if (ref_clip is None) or (moving_clip is None): return None, None, None matcher_shift_x, matcher_shift_y, metrics, corrmap = func(moving_clip, ref_clip, **kwargs) if matcher_shift_x is None: return None, None, None # Apply the shift to the center of the moving roi to the center of the reference ROI in index space. One pixel == one index (unitless). print(matcher_shift_x, matcher_shift_y) # Transformation of the ROI center if the clip applies a non-identity affine transformation affine_transformed_center_x, affine_transformed_center_y = affine((moving_roi.center[0], moving_roi.center[1]))[0] new_center_x = affine_transformed_center_x + matcher_shift_x new_center_y = affine_transformed_center_y + matcher_shift_y translation_x, translation_y = affine.inverse((new_center_x, new_center_y))[0] new_affine = tf.AffineTransform(translation=(translation_x, translation_y)) """# Apply the shift to the center of the moving roi to the center of the reference ROI in index space. One pixel == one index (unitless). new_affine_transformed_center_x = moving_roi.center[0] + matcher_shift_x #Center is indices. new_affine_transformed_center_y = moving_roi.center[1] + matcher_shift_y Loading @@ -289,10 +301,12 @@ def subpixel_template(reference_roi, # These are the inverse of the translation so that the caller can use affine() to # apply the proper translation. Otherwise, the caller would need to use affine.inverse translation_x = -(moving_roi.center[0] - inverse_transformed_affine_center_x) + moving_roi.axr translation_y = -(moving_roi.center[1] - inverse_transformed_affine_center_y) + moving_roi.ayr translation_x = -(moving_roi.center[0] - inverse_transformed_affine_center_x) translation_y = -(moving_roi.center[1] - inverse_transformed_affine_center_y)""" new_affine = tf.AffineTransform(translation=(translation_x, translation_y)) #translation_x, translation_y = affine.inverse((matcher_shift_x, matcher_shift_y))[0] #new_affine = tf.AffineTransform(translation=(translation_x, translation_y)) return new_affine, metrics, corrmap Loading autocnet/transformation/roi.py +68 −26 Original line number Diff line number Diff line from math import modf, floor import numpy as np import scipy.ndimage as ndimage from skimage import transform as tf from skimage.util import img_as_float32 Loading Loading @@ -40,30 +42,43 @@ class Roi(): bottom_y : int The bottom image coordinate in imge space """ def __init__(self, data, x, y, size_x=200, size_y=200, ndv=None, ndv_threshold=0.5): def __init__(self, data, x, y, size_x=200, size_y=200, ndv=None, ndv_threshold=0.5, buffer=5): self.data = data self.x = x self.y = y self._affine = tf.AffineTransform(translation=(self._remainder_x, self._remainder_y)) self.size_x = size_x self.size_y = size_y self.ndv = ndv self._ndv_threshold = ndv_threshold self.buffer = buffer @property def affine(self): """ This affine sets the origin of the ROI to be (0,0). """ return self._affine @property def x(self): return self._x + self.axr return self._whole_x + self._remainder_x @x.setter def x(self, x): self.axr, self._x = modf(x) self._whole_x = round(x, 0) self._remainder_x = x - self._whole_x return self._whole_x + self._remainder_x @property def y(self): return self._y + self.ayr return self._whole_y + self._remainder_y @y.setter def y(self, y): self.ayr, self._y = modf(y) self._whole_y = round(y, 0) self._remainder_y = y - self._whole_y @property def ndv_threshold(self): Loading Loading @@ -124,20 +139,21 @@ class Roi(): # Should this modify (+-) and then round to whole pixel? # what is the extent that can actually be extracted? left_x = self._x - self.size_x right_x = self._x + self.size_x top_y = self._y - self.size_y bottom_y = self._y + self.size_y left_x = self._whole_x - self.size_x right_x = self._whole_x + self.size_x top_y = self._whole_y - self.size_y bottom_y = self._whole_y + self.size_y if left_x < 0 or top_y < 0 or right_x > raster_size[0] or bottom_y > raster_size[1]: raise IndexError(f"Input window size {(self.size_x, self.size_y)}) at center {(self.x, self.y)} is out of the image bounds") #if left_x < 0 or top_y < 0 or right_x > raster_size[0] or bottom_y > raster_size[1]: # raise IndexError(f"Input window size {(self.size_x, self.size_y)}) at center {(self.x, self.y)} is out of the image bounds") return list(map(int, [left_x, right_x, top_y, bottom_y])) @property def center(self): ie = self.image_extent return ((ie[1] - ie[0])-1)/2. + 0.5, ((ie[3]-ie[2])-1)/2. + 0.5 return (0,0) #ie = self.image_extent #return ((ie[1] - ie[0])-1)/2. + 0.5, ((ie[3]-ie[2])-1)/2. + 0.5 @property def is_valid(self): Loading @@ -161,15 +177,8 @@ class Roi(): """ The clipped array associated with this ROI. """ pixels = self.image_extent if isinstance(self.data, np.ndarray): data = self.data[pixels[2]:pixels[3]+1,pixels[0]:pixels[1]+1] else: # Have to reformat to [xstart, ystart, xnumberpixels, ynumberpixels] # TODO: I think this will result in an incorrect obj.center when the passed data is a GeoDataset pixels = [pixels[0], pixels[2], pixels[1]-pixels[0]+1, pixels[3]-pixels[2]+1] data = self.data.read_array(pixels=pixels) return img_as_float32(data) return self.clip() def clip(self, affine=None, dtype=None, mode="constant"): """ Loading @@ -189,15 +198,48 @@ class Roi(): """ self.dtype = dtype pixels = self.image_extent if (np.asarray(pixels) - self.buffer < 0).any(): raise IndexError('Image coordinates plus read buffer are outside of the available data. Please select a smaller ROI and/or a smaller read buffer.') if isinstance(self.data, np.ndarray): data = self.data[pixels[2]-self.buffer:pixels[3]+1+self.buffer, pixels[0]-self.buffer:pixels[1]+1+self.buffer] else: # Have to reformat to [xstart, ystart, xnumberpixels, ynumberpixels] # TODO: I think this will result in an incorrect obj.center when the passed data is a GeoDataset pixels = [pixels[0]-self.buffer, pixels[2]-self.buffer, pixels[1]-pixels[0]+1+self.buffer, pixels[3]-pixels[2]+1+self.buffer] data = self.data.read_array(pixels=pixels) # Now that the whole pixel array has been read, interpolate the array to align pixel edges xi = np.linspace(self.buffer + self._remainder_x, self.buffer + self._remainder_x + (self.size_x*2), self.size_x*2+1) yi = np.linspace(self.buffer + self._remainder_y, self.buffer + self._remainder_y + (self.size_y*2), self.size_y*2+1) pixel_locked = ndimage.map_coordinates(data, np.meshgrid(xi, yi, indexing='ij'), mode='constant', cval=0.0, order=3) if affine: array_to_warp = self.array # if array_to_warp.shape != ((self.size_y * 2) + 1, (self.size_x * 2) + 1): # raise ValueError("Unable to enlarge Roi to apply affine transformation." + # f" Was only able to extract {array_to_warp.shape}, when " + # f"{((self.size_y * 2) + 1, (self.size_x * 2) + 1)} was asked for. Select, " + # "a smaller region of interest" ) transformed_array = tf.warp(array_to_warp, affine.inverse, order=3, mode=mode, cval=0) return transformed_array pixel_locked = tf.warp(pixel_locked, affine.inverse, order=3, mode=mode, cval=0) return img_as_float32(self.array) if self.buffer != 0: return img_as_float32(pixel_locked[self.buffer:-self.buffer, self.buffer:-self.buffer]) else: return img_as_float32(pixel_locked) Loading
autocnet/matcher/naive_template.py +2 −2 Original line number Diff line number Diff line Loading @@ -130,7 +130,7 @@ def pattern_match(template, image, upsampling=8, metric=cv2.TM_CCOEFF_NORMED, er # Pad the result array with values outside the valid correlation range width = (np.asarray(u_template.shape) - np.asarray(corrmap.shape)) // 2 print(width) result = np.pad(corrmap, width, mode='constant') # pads zeros if metric == cv2.TM_SQDIFF or metric == cv2.TM_SQDIFF_NORMED: matched_y, matched_x = np.unravel_index(np.argmin(result), result.shape) Loading
autocnet/matcher/subpixel.py +20 −6 Original line number Diff line number Diff line Loading @@ -270,11 +270,23 @@ def subpixel_template(reference_roi, if (ref_clip is None) or (moving_clip is None): return None, None, None matcher_shift_x, matcher_shift_y, metrics, corrmap = func(moving_clip, ref_clip, **kwargs) if matcher_shift_x is None: return None, None, None # Apply the shift to the center of the moving roi to the center of the reference ROI in index space. One pixel == one index (unitless). print(matcher_shift_x, matcher_shift_y) # Transformation of the ROI center if the clip applies a non-identity affine transformation affine_transformed_center_x, affine_transformed_center_y = affine((moving_roi.center[0], moving_roi.center[1]))[0] new_center_x = affine_transformed_center_x + matcher_shift_x new_center_y = affine_transformed_center_y + matcher_shift_y translation_x, translation_y = affine.inverse((new_center_x, new_center_y))[0] new_affine = tf.AffineTransform(translation=(translation_x, translation_y)) """# Apply the shift to the center of the moving roi to the center of the reference ROI in index space. One pixel == one index (unitless). new_affine_transformed_center_x = moving_roi.center[0] + matcher_shift_x #Center is indices. new_affine_transformed_center_y = moving_roi.center[1] + matcher_shift_y Loading @@ -289,10 +301,12 @@ def subpixel_template(reference_roi, # These are the inverse of the translation so that the caller can use affine() to # apply the proper translation. Otherwise, the caller would need to use affine.inverse translation_x = -(moving_roi.center[0] - inverse_transformed_affine_center_x) + moving_roi.axr translation_y = -(moving_roi.center[1] - inverse_transformed_affine_center_y) + moving_roi.ayr translation_x = -(moving_roi.center[0] - inverse_transformed_affine_center_x) translation_y = -(moving_roi.center[1] - inverse_transformed_affine_center_y)""" new_affine = tf.AffineTransform(translation=(translation_x, translation_y)) #translation_x, translation_y = affine.inverse((matcher_shift_x, matcher_shift_y))[0] #new_affine = tf.AffineTransform(translation=(translation_x, translation_y)) return new_affine, metrics, corrmap Loading
autocnet/transformation/roi.py +68 −26 Original line number Diff line number Diff line from math import modf, floor import numpy as np import scipy.ndimage as ndimage from skimage import transform as tf from skimage.util import img_as_float32 Loading Loading @@ -40,30 +42,43 @@ class Roi(): bottom_y : int The bottom image coordinate in imge space """ def __init__(self, data, x, y, size_x=200, size_y=200, ndv=None, ndv_threshold=0.5): def __init__(self, data, x, y, size_x=200, size_y=200, ndv=None, ndv_threshold=0.5, buffer=5): self.data = data self.x = x self.y = y self._affine = tf.AffineTransform(translation=(self._remainder_x, self._remainder_y)) self.size_x = size_x self.size_y = size_y self.ndv = ndv self._ndv_threshold = ndv_threshold self.buffer = buffer @property def affine(self): """ This affine sets the origin of the ROI to be (0,0). """ return self._affine @property def x(self): return self._x + self.axr return self._whole_x + self._remainder_x @x.setter def x(self, x): self.axr, self._x = modf(x) self._whole_x = round(x, 0) self._remainder_x = x - self._whole_x return self._whole_x + self._remainder_x @property def y(self): return self._y + self.ayr return self._whole_y + self._remainder_y @y.setter def y(self, y): self.ayr, self._y = modf(y) self._whole_y = round(y, 0) self._remainder_y = y - self._whole_y @property def ndv_threshold(self): Loading Loading @@ -124,20 +139,21 @@ class Roi(): # Should this modify (+-) and then round to whole pixel? # what is the extent that can actually be extracted? left_x = self._x - self.size_x right_x = self._x + self.size_x top_y = self._y - self.size_y bottom_y = self._y + self.size_y left_x = self._whole_x - self.size_x right_x = self._whole_x + self.size_x top_y = self._whole_y - self.size_y bottom_y = self._whole_y + self.size_y if left_x < 0 or top_y < 0 or right_x > raster_size[0] or bottom_y > raster_size[1]: raise IndexError(f"Input window size {(self.size_x, self.size_y)}) at center {(self.x, self.y)} is out of the image bounds") #if left_x < 0 or top_y < 0 or right_x > raster_size[0] or bottom_y > raster_size[1]: # raise IndexError(f"Input window size {(self.size_x, self.size_y)}) at center {(self.x, self.y)} is out of the image bounds") return list(map(int, [left_x, right_x, top_y, bottom_y])) @property def center(self): ie = self.image_extent return ((ie[1] - ie[0])-1)/2. + 0.5, ((ie[3]-ie[2])-1)/2. + 0.5 return (0,0) #ie = self.image_extent #return ((ie[1] - ie[0])-1)/2. + 0.5, ((ie[3]-ie[2])-1)/2. + 0.5 @property def is_valid(self): Loading @@ -161,15 +177,8 @@ class Roi(): """ The clipped array associated with this ROI. """ pixels = self.image_extent if isinstance(self.data, np.ndarray): data = self.data[pixels[2]:pixels[3]+1,pixels[0]:pixels[1]+1] else: # Have to reformat to [xstart, ystart, xnumberpixels, ynumberpixels] # TODO: I think this will result in an incorrect obj.center when the passed data is a GeoDataset pixels = [pixels[0], pixels[2], pixels[1]-pixels[0]+1, pixels[3]-pixels[2]+1] data = self.data.read_array(pixels=pixels) return img_as_float32(data) return self.clip() def clip(self, affine=None, dtype=None, mode="constant"): """ Loading @@ -189,15 +198,48 @@ class Roi(): """ self.dtype = dtype pixels = self.image_extent if (np.asarray(pixels) - self.buffer < 0).any(): raise IndexError('Image coordinates plus read buffer are outside of the available data. Please select a smaller ROI and/or a smaller read buffer.') if isinstance(self.data, np.ndarray): data = self.data[pixels[2]-self.buffer:pixels[3]+1+self.buffer, pixels[0]-self.buffer:pixels[1]+1+self.buffer] else: # Have to reformat to [xstart, ystart, xnumberpixels, ynumberpixels] # TODO: I think this will result in an incorrect obj.center when the passed data is a GeoDataset pixels = [pixels[0]-self.buffer, pixels[2]-self.buffer, pixels[1]-pixels[0]+1+self.buffer, pixels[3]-pixels[2]+1+self.buffer] data = self.data.read_array(pixels=pixels) # Now that the whole pixel array has been read, interpolate the array to align pixel edges xi = np.linspace(self.buffer + self._remainder_x, self.buffer + self._remainder_x + (self.size_x*2), self.size_x*2+1) yi = np.linspace(self.buffer + self._remainder_y, self.buffer + self._remainder_y + (self.size_y*2), self.size_y*2+1) pixel_locked = ndimage.map_coordinates(data, np.meshgrid(xi, yi, indexing='ij'), mode='constant', cval=0.0, order=3) if affine: array_to_warp = self.array # if array_to_warp.shape != ((self.size_y * 2) + 1, (self.size_x * 2) + 1): # raise ValueError("Unable to enlarge Roi to apply affine transformation." + # f" Was only able to extract {array_to_warp.shape}, when " + # f"{((self.size_y * 2) + 1, (self.size_x * 2) + 1)} was asked for. Select, " + # "a smaller region of interest" ) transformed_array = tf.warp(array_to_warp, affine.inverse, order=3, mode=mode, cval=0) return transformed_array pixel_locked = tf.warp(pixel_locked, affine.inverse, order=3, mode=mode, cval=0) return img_as_float32(self.array) if self.buffer != 0: return img_as_float32(pixel_locked[self.buffer:-self.buffer, self.buffer:-self.buffer]) else: return img_as_float32(pixel_locked)
