Commit 5435f82f authored by Jay's avatar Jay Committed by jay
Browse files

Backed the downsampling back out of the code. Downsampling does not work well currently.

parent efc569c7

autocnet/graph/network.py

+10 −12
Original line number Diff line number Diff line
@@ -165,7 +165,7 @@ class CandidateGraph(nx.Graph): 
        """

        self.node[nodeindex]['handle'] = GeoDataset(self.node[nodeindex]['image_path'])



    def get_array(self, nodeindex, downsampling=1):

    def get_array(self, nodeindex):

        """

        Downsample the input image file by some amount using bicubic interpolation

        in order to reduce data sizes for visualization and analysis, e.g. feature detection
@@ -180,10 +180,9 @@ class CandidateGraph(nx.Graph): 
        """



        array = self.node[nodeindex]['handle'].read_array()

        newx_size = int(array.shape[0] / downsampling)

        newy_size = int(array.shape[1] / downsampling)

        newx_size = int(array.shape[0])

        newy_size = int(array.shape[1])



        resized_array = imresize(array, (newx_size, newy_size), interp='bicubic')

        return bytescale(resized_array)



    def extract_features(self, method='orb', extractor_parameters={}, downsampling=1):
@@ -204,8 +203,7 @@ class CandidateGraph(nx.Graph): 
        """

        for node, attributes in self.nodes_iter(data=True):

            self.get_geodataset(node)

            attributes['downsampling'] = downsampling

            image = self.get_array(node, downsampling=downsampling)

            image = self.get_array(node)

            keypoints, descriptors = fe.extract_features(image,

                                                         method=method,

                                                         extractor_parameters=extractor_parameters)
@@ -358,15 +356,15 @@ class CandidateGraph(nx.Graph): 
                s_idx = int(row['source_idx'])

                d_idx = int(row['destination_idx'])



                s_keypoint = [s_node['keypoints'][s_idx].pt[0] * s_node['downsampling'],

                              s_node['keypoints'][s_idx].pt[1] * s_node['downsampling']]

                s_keypoint = [s_node['keypoints'][s_idx].pt[0],

                              s_node['keypoints'][s_idx].pt[1]]



                d_keypoint = [d_node['keypoints'][d_idx].pt[0] * d_node['downsampling'],

                              d_node['keypoints'][d_idx].pt[1] * d_node['downsampling']]

                d_keypoint = [d_node['keypoints'][d_idx].pt[0],

                              d_node['keypoints'][d_idx].pt[1]]



                # Get the template and search windows

                s_template = sp.clip_roi(s_image, s_keypoint, template_size * s_node['downsampling'])

                d_search = sp.clip_roi(d_image, d_keypoint, search_size * d_node['downsampling'])

                s_template = sp.clip_roi(s_image, s_keypoint, template_size)

                d_search = sp.clip_roi(d_image, d_keypoint, search_size)



                edge_offsets[i] = sp.subpixel_offset(s_template, d_search, upsampling=upsampling)