coord_calc.py

# -*- coding: utf-8 -*-

# py_tools_ds - A collection of geospatial data analysis tools that simplify standard
# operations when handling geospatial raster and vector data as well as projections.
#
# Copyright (C) 2016-2021
# - Daniel Scheffler (GFZ Potsdam, daniel.scheffler@gfz-potsdam.de)
# - Helmholtz Centre Potsdam - GFZ German Research Centre for Geosciences Potsdam,
#   Germany (https://www.gfz-potsdam.de/)
#
# This software was developed within the context of the GeoMultiSens project funded
# by the German Federal Ministry of Education and Research
# (project grant code: 01 IS 14 010 A-C).
#
# Licensed under the Apache License, Version 2.0 (the "License");
# you may not use this file except in compliance with the License.
# You may obtain a copy of the License at
#
#   http://www.apache.org/licenses/LICENSE-2.0
#
# Unless required by applicable law or agreed to in writing, software
# distributed under the License is distributed on an "AS IS" BASIS,
# WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
# See the License for the specific language governing permissions and
# limitations under the License.

import warnings
from typing import Iterable, Tuple  # noqa: F401
import numpy as np

# custom
from shapely.geometry import Polygon

__author__ = "Daniel Scheffler"


def get_corner_coordinates(gt, cols, rows):
    """Return (ULxy, LLxy, LRxy, URxy) in the same coordinate units like the given geotransform."""
    ext = []
    xarr = [0, cols]
    yarr = [0, rows]
    for px in xarr:
        for py in yarr:
            x = gt[0] + (px * gt[1]) + (py * gt[2])
            y = gt[3] + (px * gt[4]) + (py * gt[5])
            ext.append([x, y])
        yarr.reverse()
    return ext


def calc_FullDataset_corner_positions(mask_1bit, assert_four_corners=True, algorithm='shapely'):
    # type: (np.ndarray, bool, str) -> list
    """Calculate the image coordinates of the true data corners from a nodata mask.

    NOTE: Algorithm 'shapely' calculates the corner coordinates of the convex hull of the given mask. Since the convex
    hull not always reflects all of the true corner coordinates the result can have a limitation in this regard.

    :param mask_1bit:               2D-numpy array 1bit
    :param assert_four_corners:     <bool> whether to assert four corners or not
    :param algorithm:               <str> 'shapely' or 'numpy' (default: 'shapely')
    :return:                        [UL, UR, LL, LR] as [(ULrow,ULcol),(URrow,URcol),...]
    """
    # check if the mask extent covers real data or only nodata
    pixVals = np.unique(mask_1bit)
    if not (True in pixVals or 1 in pixVals):
        # 'According to the given mask the mask extent is completely outside the image.
        # No calculation of image corner coordinates possible.'
        return []

    rows, cols = mask_1bit.shape[:2]

    if sum([mask_1bit[0, 0], mask_1bit[0, cols - 1], mask_1bit[rows - 1, 0], mask_1bit[rows - 1, cols - 1]]) == 4:
        # if mask contains pixel value 1 in all of the four corners: return outer corner coords
        out = [(0, 0), (0, cols - 1), (rows - 1, 0), (rows - 1, cols - 1)]  # UL, UR, LL, LR
    else:
        assert algorithm in ['shapely', 'numpy'], "Algorithm '%' does not exist. Choose between 'shapely' and 'numpy'."

        if algorithm == 'shapely':
            # get outline
            topbottom = np.empty((1, 2 * mask_1bit.shape[1]), dtype=np.uint16)
            topbottom[0, 0:mask_1bit.shape[1]] = np.argmax(mask_1bit, axis=0)
            topbottom[0, mask_1bit.shape[1]:] = (mask_1bit.shape[0] - 1) - np.argmax(np.flipud(mask_1bit), axis=0)
            mask = np.tile(np.any(mask_1bit, axis=0), (2,))
            xvalues = np.tile(np.arange(mask_1bit.shape[1]), (1, 2))
            outline = np.vstack([topbottom, xvalues])[:, mask].T

            with warnings.catch_warnings():
                warnings.simplefilter('ignore')  # FIXME not working
                poly = Polygon(outline).convex_hull
                poly = poly.simplify(20, preserve_topology=True)
                poly = Polygon(list(set(
                    poly.exterior.coords))).convex_hull  # eliminiate duplicates except of 1 (because poly is closed)
                unique_corn_YX = sorted(set(poly.exterior.coords),
                                        key=lambda x: poly.exterior.coords[:].index(x))  # [(rows,cols),rows,cols]

            if assert_four_corners or len(unique_corn_YX) == 4:
                # sort calculated corners like this: [UL, UR, LL, LR]
                assert len(unique_corn_YX) == 4, \
                    'Found %s unique corner coordinates instead of 4. Exiting.' % len(unique_corn_YX)

                def get_dist(P1yx, P2yx):
                    return np.sqrt((P1yx[0] - P2yx[0]) ** 2 + (P1yx[1] - P2yx[1]) ** 2)

                def getClosest(tgtYX, srcYXs):
                    return srcYXs[np.array([get_dist(tgtYX, srcYX) for srcYX in srcYXs]).argmin()]

                out = [getClosest(tgtYX, unique_corn_YX) for tgtYX in
                       [(0, 0), (0, cols - 1), (rows - 1, 0), (rows - 1, cols - 1)]]
            else:
                out = unique_corn_YX

        else:  # alg='numpy'
            cols_containing_data = np.any(mask_1bit, axis=0)
            rows_containing_data = np.any(mask_1bit, axis=1)

            first_dataCol = list(cols_containing_data).index(True)
            last_dataCol = cols - (list(reversed(cols_containing_data)).index(True) + 1)
            first_dataRow = list(rows_containing_data).index(True)
            last_dataRow = rows - (list(reversed(rows_containing_data)).index(True) + 1)

            StartStopRows_in_first_dataCol = [list(mask_1bit[:, first_dataCol]).index(True),
                                              (rows - list(reversed(mask_1bit[:, first_dataCol])).index(True) + 1)]
            StartStopRows_in_last_dataCol = [list(mask_1bit[:, last_dataCol]).index(True),
                                             (rows - list(reversed(mask_1bit[:, last_dataCol])).index(True) + 1)]
            StartStopCols_in_first_dataRow = [list(mask_1bit[first_dataRow, :]).index(True),
                                              (cols - list(reversed(mask_1bit[first_dataRow, :])).index(True) + 1)]
            StartStopCols_in_last_dataRow = [list(mask_1bit[last_dataRow, :]).index(True),
                                             (cols - list(reversed(mask_1bit[last_dataRow, :])).index(True) + 1)]

            if True in [abs(np.diff(i)[0]) > 10 for i in
                        [StartStopRows_in_first_dataCol, StartStopRows_in_last_dataCol,
                         StartStopCols_in_first_dataRow, StartStopCols_in_last_dataRow]]:
                # In case of cut image corners (e.g. ALOS AVNIR-2 Level-1B2):
                # Calculation of trueDataCornerPos outside of the image.'''
                line_N = ((StartStopCols_in_first_dataRow[1], first_dataRow),
                          (last_dataCol, StartStopRows_in_last_dataCol[0]))
                line_S = ((first_dataCol, StartStopRows_in_first_dataCol[1]),
                          (StartStopCols_in_last_dataRow[0], last_dataRow))
                line_W = ((StartStopCols_in_first_dataRow[0], first_dataRow),
                          (first_dataCol, StartStopRows_in_first_dataCol[0]))
                line_O = ((last_dataCol, StartStopRows_in_last_dataCol[1]),
                          (StartStopCols_in_last_dataRow[1], last_dataRow))
                from .vector.topology import find_line_intersection_point  # import here avoids circular dependencies
                corners = [list(reversed(find_line_intersection_point(line_N, line_W))),
                           list(reversed(find_line_intersection_point(line_N, line_O))),
                           list(reversed(find_line_intersection_point(line_S, line_W))),
                           list(reversed(find_line_intersection_point(line_S, line_O)))]
            else:
                dataRow_in_first_dataCol = np.mean(StartStopRows_in_first_dataCol)
                dataRow_in_last_dataCol = np.mean(StartStopRows_in_last_dataCol)
                dataCol_in_first_dataRow = np.mean(StartStopCols_in_first_dataRow)
                dataCol_in_last_dataRow = np.mean(StartStopCols_in_last_dataRow)
                corners = [(first_dataRow, dataCol_in_first_dataRow), (dataRow_in_last_dataCol, last_dataCol),
                           (last_dataRow, dataCol_in_last_dataRow), (dataRow_in_first_dataCol, first_dataCol)]

            def getClosest(refR, refC):
                return corners[np.array([np.sqrt((refR - c[0]) ** 2 + (refC - c[1]) ** 2) for c in
                                         corners]).argmin()]  # FIXME this can also result in only 2 corners
            out = [getClosest(refR, refC) for refR, refC in
                   [(0, 0), (0, cols - 1), (rows - 1, 0), (rows - 1, cols - 1)]]
            # if UL[0] == 0 and UR[0] == 0:
            #     envi.save_image(os.path.abspath('./testing/out/__faulty_mask_1bit.hdr'),
            #                     self.mask_1bit, dtype= 'uint8', interleave='bsq', ext='.bsq',force=True)
            # print(UL,UR,LL,LR)
    return out


def corner_coord_to_minmax(corner_coords):
    # type: (Iterable[Tuple[float, float], Tuple[float, float], Tuple[float, float], Tuple[float, float]]) -> tuple
    """Return the bounding coordinates for a given set of XY coordinates.

    :param corner_coords:   # four XY tuples of corner coordinates. Their order does not matter.
    :return: xmin,xmax,ymin,ymax
    """
    x_vals = [i[0] for i in corner_coords]
    y_vals = [i[1] for i in corner_coords]
    xmin, xmax, ymin, ymax = min(x_vals), max(x_vals), min(y_vals), max(y_vals)

    return xmin, xmax, ymin, ymax