testbed/astropy__astropy/astropy/convolution/tests/test_kernel_class.py

# Licensed under a 3-clause BSD style license - see LICENSE.rst

import itertools

import pytest
import numpy as np
from numpy.testing import assert_almost_equal, assert_allclose

from astropy.convolution.convolve import convolve, convolve_fft
from astropy.convolution.kernels import (
    Gaussian1DKernel, Gaussian2DKernel, Box1DKernel, Box2DKernel,
    Trapezoid1DKernel, TrapezoidDisk2DKernel, MexicanHat1DKernel,
    Tophat2DKernel, MexicanHat2DKernel, AiryDisk2DKernel, Ring2DKernel,
    CustomKernel, Model1DKernel, Model2DKernel, Kernel1D, Kernel2D)

from astropy.convolution.utils import KernelSizeError
from astropy.modeling.models import Box2D, Gaussian1D, Gaussian2D
from astropy.utils.exceptions import AstropyDeprecationWarning, AstropyUserWarning

try:
    from scipy.ndimage import filters
    HAS_SCIPY = True
except ImportError:
    HAS_SCIPY = False

WIDTHS_ODD = [3, 5, 7, 9]
WIDTHS_EVEN = [2, 4, 8, 16]
MODES = ['center', 'linear_interp', 'oversample', 'integrate']
KERNEL_TYPES = [Gaussian1DKernel, Gaussian2DKernel,
                Box1DKernel, Box2DKernel,
                Trapezoid1DKernel, TrapezoidDisk2DKernel,
                MexicanHat1DKernel, Tophat2DKernel, AiryDisk2DKernel,
                Ring2DKernel]


NUMS = [1, 1., np.float32(1.), np.float64(1.)]


# Test data
delta_pulse_1D = np.zeros(81)
delta_pulse_1D[40] = 1

delta_pulse_2D = np.zeros((81, 81))
delta_pulse_2D[40, 40] = 1

random_data_1D = np.random.rand(61)
random_data_2D = np.random.rand(61, 61)


class TestKernels:
    """
    Test class for the built-in convolution kernels.
    """

    @pytest.mark.skipif('not HAS_SCIPY')
    @pytest.mark.parametrize(('width'), WIDTHS_ODD)
    def test_scipy_filter_gaussian(self, width):
        """
        Test GaussianKernel against SciPy ndimage gaussian filter.
        """
        gauss_kernel_1D = Gaussian1DKernel(width)
        gauss_kernel_1D.normalize()
        gauss_kernel_2D = Gaussian2DKernel(width)
        gauss_kernel_2D.normalize()

        astropy_1D = convolve(delta_pulse_1D, gauss_kernel_1D, boundary='fill')
        astropy_2D = convolve(delta_pulse_2D, gauss_kernel_2D, boundary='fill')

        scipy_1D = filters.gaussian_filter(delta_pulse_1D, width)
        scipy_2D = filters.gaussian_filter(delta_pulse_2D, width)

        assert_almost_equal(astropy_1D, scipy_1D, decimal=12)
        assert_almost_equal(astropy_2D, scipy_2D, decimal=12)

    @pytest.mark.skipif('not HAS_SCIPY')
    @pytest.mark.parametrize(('width'), WIDTHS_ODD)
    def test_scipy_filter_gaussian_laplace(self, width):
        """
        Test MexicanHat kernels against SciPy ndimage gaussian laplace filters.
        """
        mexican_kernel_1D = MexicanHat1DKernel(width)
        mexican_kernel_2D = MexicanHat2DKernel(width)

        astropy_1D = convolve(delta_pulse_1D, mexican_kernel_1D, boundary='fill', normalize_kernel=False)
        astropy_2D = convolve(delta_pulse_2D, mexican_kernel_2D, boundary='fill', normalize_kernel=False)

        with pytest.raises(Exception) as exc:
            astropy_1D = convolve(delta_pulse_1D, mexican_kernel_1D, boundary='fill', normalize_kernel=True)
        assert 'sum is close to zero' in exc.value.args[0]

        with pytest.raises(Exception) as exc:
            astropy_2D = convolve(delta_pulse_2D, mexican_kernel_2D, boundary='fill', normalize_kernel=True)
        assert 'sum is close to zero' in exc.value.args[0]

        # The Laplace of Gaussian filter is an inverted Mexican Hat
        # filter.
        scipy_1D = -filters.gaussian_laplace(delta_pulse_1D, width)
        scipy_2D = -filters.gaussian_laplace(delta_pulse_2D, width)

        # There is a slight deviation in the normalization. They differ by a
        # factor of ~1.0000284132604045. The reason is not known.
        assert_almost_equal(astropy_1D, scipy_1D, decimal=5)
        assert_almost_equal(astropy_2D, scipy_2D, decimal=5)

    @pytest.mark.parametrize(('kernel_type', 'width'), list(itertools.product(KERNEL_TYPES, WIDTHS_ODD)))
    def test_delta_data(self, kernel_type, width):
        """
        Test smoothing of an image with a single positive pixel
        """
        if kernel_type == AiryDisk2DKernel and not HAS_SCIPY:
            pytest.skip("Omitting AiryDisk2DKernel, which requires SciPy")
        if not kernel_type == Ring2DKernel:
            kernel = kernel_type(width)
        else:
            kernel = kernel_type(width, width * 0.2)

        if kernel.dimension == 1:
            c1 = convolve_fft(delta_pulse_1D, kernel, boundary='fill', normalize_kernel=False)
            c2 = convolve(delta_pulse_1D, kernel, boundary='fill', normalize_kernel=False)
            assert_almost_equal(c1, c2, decimal=12)
        else:
            c1 = convolve_fft(delta_pulse_2D, kernel, boundary='fill', normalize_kernel=False)
            c2 = convolve(delta_pulse_2D, kernel, boundary='fill', normalize_kernel=False)
            assert_almost_equal(c1, c2, decimal=12)

    @pytest.mark.parametrize(('kernel_type', 'width'), list(itertools.product(KERNEL_TYPES, WIDTHS_ODD)))
    def test_random_data(self, kernel_type, width):
        """
        Test smoothing of an image made of random noise
        """
        if kernel_type == AiryDisk2DKernel and not HAS_SCIPY:
            pytest.skip("Omitting AiryDisk2DKernel, which requires SciPy")
        if not kernel_type == Ring2DKernel:
            kernel = kernel_type(width)
        else:
            kernel = kernel_type(width, width * 0.2)

        if kernel.dimension == 1:
            c1 = convolve_fft(random_data_1D, kernel, boundary='fill', normalize_kernel=False)
            c2 = convolve(random_data_1D, kernel, boundary='fill', normalize_kernel=False)
            assert_almost_equal(c1, c2, decimal=12)
        else:
            c1 = convolve_fft(random_data_2D, kernel, boundary='fill', normalize_kernel=False)
            c2 = convolve(random_data_2D, kernel, boundary='fill', normalize_kernel=False)
            assert_almost_equal(c1, c2, decimal=12)

    @pytest.mark.parametrize(('width'), WIDTHS_ODD)
    def test_uniform_smallkernel(self, width):
        """
        Test smoothing of an image with a single positive pixel

        Instead of using kernel class, uses a simple, small kernel
        """
        kernel = np.ones([width, width])

        c2 = convolve_fft(delta_pulse_2D, kernel, boundary='fill')
        c1 = convolve(delta_pulse_2D, kernel, boundary='fill')
        assert_almost_equal(c1, c2, decimal=12)

    @pytest.mark.parametrize(('width'), WIDTHS_ODD)
    def test_smallkernel_vs_Box2DKernel(self, width):
        """
        Test smoothing of an image with a single positive pixel
        """
        kernel1 = np.ones([width, width]) / width ** 2
        kernel2 = Box2DKernel(width)

        c2 = convolve_fft(delta_pulse_2D, kernel2, boundary='fill')
        c1 = convolve_fft(delta_pulse_2D, kernel1, boundary='fill')

        assert_almost_equal(c1, c2, decimal=12)

    def test_convolve_1D_kernels(self):
        """
        Check if convolving two kernels with each other works correctly.
        """
        gauss_1 = Gaussian1DKernel(3)
        gauss_2 = Gaussian1DKernel(4)
        test_gauss_3 = Gaussian1DKernel(5)

        with pytest.warns(AstropyUserWarning, match='Both array and kernel '
                          'are Kernel instances'):
            gauss_3 = convolve(gauss_1, gauss_2)

        assert np.all(np.abs((gauss_3 - test_gauss_3).array) < 0.01)

    def test_convolve_2D_kernels(self):
        """
        Check if convolving two kernels with each other works correctly.
        """
        gauss_1 = Gaussian2DKernel(3)
        gauss_2 = Gaussian2DKernel(4)
        test_gauss_3 = Gaussian2DKernel(5)

        with pytest.warns(AstropyUserWarning, match='Both array and kernel '
                          'are Kernel instances'):
            gauss_3 = convolve(gauss_1, gauss_2)

        assert np.all(np.abs((gauss_3 - test_gauss_3).array) < 0.01)

    @pytest.mark.parametrize(('number'), NUMS)
    def test_multiply_scalar(self, number):
        """
        Check if multiplying a kernel with a scalar works correctly.
        """
        gauss = Gaussian1DKernel(3)
        gauss_new = number * gauss
        assert_almost_equal(gauss_new.array, gauss.array * number, decimal=12)

    @pytest.mark.parametrize(('number'), NUMS)
    def test_multiply_scalar_type(self, number):
        """
        Check if multiplying a kernel with a scalar works correctly.
        """
        gauss = Gaussian1DKernel(3)
        gauss_new = number * gauss
        assert type(gauss_new) is Gaussian1DKernel

    @pytest.mark.parametrize(('number'), NUMS)
    def test_rmultiply_scalar_type(self, number):
        """
        Check if multiplying a kernel with a scalar works correctly.
        """
        gauss = Gaussian1DKernel(3)
        gauss_new = gauss * number
        assert type(gauss_new) is Gaussian1DKernel

    def test_multiply_kernel1d(self):
        """Test that multiplying two 1D kernels raises an exception."""
        gauss = Gaussian1DKernel(3)
        with pytest.raises(Exception):
            gauss * gauss

    def test_multiply_kernel2d(self):
        """Test that multiplying two 2D kernels raises an exception."""
        gauss = Gaussian2DKernel(3)
        with pytest.raises(Exception):
            gauss * gauss

    def test_multiply_kernel1d_kernel2d(self):
        """
        Test that multiplying a 1D kernel with a 2D kernel raises an
        exception.
        """
        with pytest.raises(Exception):
            Gaussian1DKernel(3) * Gaussian2DKernel(3)

    def test_add_kernel_scalar(self):
        """Test that adding a scalar to a kernel raises an exception."""
        with pytest.raises(Exception):
            Gaussian1DKernel(3) + 1

    def test_model_1D_kernel(self):
        """
        Check Model1DKernel against Gaussian1Dkernel
        """
        stddev = 5.
        gauss = Gaussian1D(1. / np.sqrt(2 * np.pi * stddev**2), 0, stddev)
        model_gauss_kernel = Model1DKernel(gauss, x_size=21)
        gauss_kernel = Gaussian1DKernel(stddev, x_size=21)
        assert_almost_equal(model_gauss_kernel.array, gauss_kernel.array,
                            decimal=12)

    def test_model_2D_kernel(self):
        """
        Check Model2DKernel against Gaussian2Dkernel
        """
        stddev = 5.
        gauss = Gaussian2D(1. / (2 * np.pi * stddev**2), 0, 0, stddev, stddev)
        model_gauss_kernel = Model2DKernel(gauss, x_size=21)
        gauss_kernel = Gaussian2DKernel(stddev, x_size=21)
        assert_almost_equal(model_gauss_kernel.array, gauss_kernel.array,
                            decimal=12)

    def test_custom_1D_kernel(self):
        """
        Check CustomKernel against Box1DKernel.
        """
        # Define one dimensional array:
        array = np.ones(5)
        custom = CustomKernel(array)
        custom.normalize()
        box = Box1DKernel(5)

        c2 = convolve(delta_pulse_1D, custom, boundary='fill')
        c1 = convolve(delta_pulse_1D, box, boundary='fill')
        assert_almost_equal(c1, c2, decimal=12)

    def test_custom_2D_kernel(self):
        """
        Check CustomKernel against Box2DKernel.
        """
        # Define one dimensional array:
        array = np.ones((5, 5))
        custom = CustomKernel(array)
        custom.normalize()
        box = Box2DKernel(5)

        c2 = convolve(delta_pulse_2D, custom, boundary='fill')
        c1 = convolve(delta_pulse_2D, box, boundary='fill')
        assert_almost_equal(c1, c2, decimal=12)

    def test_custom_1D_kernel_list(self):
        """
        Check if CustomKernel works with lists.
        """
        custom = CustomKernel([1, 1, 1, 1, 1])
        assert custom.is_bool is True

    def test_custom_2D_kernel_list(self):
        """
        Check if CustomKernel works with lists.
        """
        custom = CustomKernel([[1, 1, 1],
                               [1, 1, 1],
                               [1, 1, 1]])
        assert custom.is_bool is True

    def test_custom_1D_kernel_zerosum(self):
        """
        Check if CustomKernel works when the input array/list
        sums to zero.
        """
        array = [-2, -1, 0, 1, 2]

        custom = CustomKernel(array)

        with pytest.warns(AstropyUserWarning, match='kernel cannot be '
                          'normalized because it sums to zero'):
            custom.normalize()

        assert custom.truncation == 0.
        assert custom._kernel_sum == 0.

    def test_custom_2D_kernel_zerosum(self):
        """
        Check if CustomKernel works when the input array/list
        sums to zero.
        """
        array = [[0, -1, 0], [-1, 4, -1], [0, -1, 0]]

        custom = CustomKernel(array)

        with pytest.warns(AstropyUserWarning, match='kernel cannot be '
                          'normalized because it sums to zero'):
            custom.normalize()

        assert custom.truncation == 0.
        assert custom._kernel_sum == 0.

    def test_custom_kernel_odd_error(self):
        """
        Check if CustomKernel raises if the array size is odd.
        """
        with pytest.raises(KernelSizeError):
            CustomKernel([1, 1, 1, 1])

    def test_add_1D_kernels(self):
        """
        Check if adding of two 1D kernels works.
        """
        box_1 = Box1DKernel(5)
        box_2 = Box1DKernel(3)
        box_3 = Box1DKernel(1)
        box_sum_1 = box_1 + box_2 + box_3
        box_sum_2 = box_2 + box_3 + box_1
        box_sum_3 = box_3 + box_1 + box_2
        ref = [1/5., 1/5. + 1/3., 1 + 1/3. + 1/5., 1/5. + 1/3., 1/5.]
        assert_almost_equal(box_sum_1.array, ref, decimal=12)
        assert_almost_equal(box_sum_2.array, ref, decimal=12)
        assert_almost_equal(box_sum_3.array, ref, decimal=12)

        # Assert that the kernels haven't changed
        assert_almost_equal(box_1.array, [0.2, 0.2, 0.2, 0.2, 0.2], decimal=12)
        assert_almost_equal(box_2.array, [1/3., 1/3., 1/3.], decimal=12)
        assert_almost_equal(box_3.array, [1], decimal=12)

    def test_add_2D_kernels(self):
        """
        Check if adding of two 1D kernels works.
        """
        box_1 = Box2DKernel(3)
        box_2 = Box2DKernel(1)
        box_sum_1 = box_1 + box_2
        box_sum_2 = box_2 + box_1
        ref = [[1 / 9., 1 / 9., 1 / 9.],
               [1 / 9., 1 + 1 / 9., 1 / 9.],
               [1 / 9., 1 / 9., 1 / 9.]]
        ref_1 = [[1 / 9., 1 / 9., 1 / 9.],
                 [1 / 9., 1 / 9., 1 / 9.],
                 [1 / 9., 1 / 9., 1 / 9.]]
        assert_almost_equal(box_2.array, [[1]], decimal=12)
        assert_almost_equal(box_1.array, ref_1, decimal=12)
        assert_almost_equal(box_sum_1.array, ref, decimal=12)
        assert_almost_equal(box_sum_2.array, ref, decimal=12)

    def test_Gaussian1DKernel_even_size(self):
        """
        Check if even size for GaussianKernel works.
        """
        gauss = Gaussian1DKernel(3, x_size=10)
        assert gauss.array.size == 10

    def test_Gaussian2DKernel_even_size(self):
        """
        Check if even size for GaussianKernel works.
        """
        gauss = Gaussian2DKernel(3, x_size=10, y_size=10)
        assert gauss.array.shape == (10, 10)

    # https://github.com/astropy/astropy/issues/3605
    def test_Gaussian2DKernel_rotated(self):
        with pytest.warns(AstropyDeprecationWarning) as w:
            Gaussian2DKernel(stddev=10)
        assert len(w) == 1

        gauss = Gaussian2DKernel(
            x_stddev=3, y_stddev=1.5, theta=0.7853981633974483,
            x_size=5, y_size=5)  # rotated 45 deg ccw
        ans = [[0.02267712, 0.02464785, 0.02029238, 0.01265463, 0.00597762],
               [0.02464785, 0.03164847, 0.03078144, 0.02267712, 0.01265463],
               [0.02029238, 0.03078144, 0.03536777, 0.03078144, 0.02029238],
               [0.01265463, 0.02267712, 0.03078144, 0.03164847, 0.02464785],
               [0.00597762, 0.01265463, 0.02029238, 0.02464785, 0.02267712]]
        assert_allclose(gauss, ans, rtol=0.001)  # Rough comparison at 0.1 %

    def test_normalize_peak(self):
        """
        Check if normalize works with peak mode.
        """
        custom = CustomKernel([1, 2, 3, 2, 1])
        custom.normalize(mode='peak')
        assert custom.array.max() == 1

    def test_check_kernel_attributes(self):
        """
        Check if kernel attributes are correct.
        """
        box = Box2DKernel(5)

        # Check truncation
        assert box.truncation == 0

        # Check model
        assert isinstance(box.model, Box2D)

        # Check center
        assert box.center == [2, 2]

        # Check normalization
        box.normalize()
        assert_almost_equal(box._kernel_sum, 1., decimal=12)

        # Check separability
        assert box.separable

    @pytest.mark.parametrize(('kernel_type', 'mode'), list(itertools.product(KERNEL_TYPES, MODES)))
    def test_discretize_modes(self, kernel_type, mode):
        """
        Check if the different modes result in kernels that work with convolve.
        Use only small kernel width, to make the test pass quickly.
        """
        if kernel_type == AiryDisk2DKernel and not HAS_SCIPY:
            pytest.skip("Omitting AiryDisk2DKernel, which requires SciPy")
        if not kernel_type == Ring2DKernel:
            kernel = kernel_type(3)
        else:
            kernel = kernel_type(3, 3 * 0.2)

        if kernel.dimension == 1:
            c1 = convolve_fft(delta_pulse_1D, kernel, boundary='fill', normalize_kernel=False)
            c2 = convolve(delta_pulse_1D, kernel, boundary='fill', normalize_kernel=False)
            assert_almost_equal(c1, c2, decimal=12)
        else:
            c1 = convolve_fft(delta_pulse_2D, kernel, boundary='fill', normalize_kernel=False)
            c2 = convolve(delta_pulse_2D, kernel, boundary='fill', normalize_kernel=False)
            assert_almost_equal(c1, c2, decimal=12)

    @pytest.mark.parametrize(('width'), WIDTHS_EVEN)
    def test_box_kernels_even_size(self, width):
        """
        Check if BoxKernel work properly with even sizes.
        """
        kernel_1D = Box1DKernel(width)
        assert kernel_1D.shape[0] % 2 != 0
        assert kernel_1D.array.sum() == 1.

        kernel_2D = Box2DKernel(width)
        assert np.all([_ % 2 != 0 for _ in kernel_2D.shape])
        assert kernel_2D.array.sum() == 1.

    def test_kernel_normalization(self):
        """
        Test that repeated normalizations do not change the kernel [#3747].
        """

        kernel = CustomKernel(np.ones(5))
        kernel.normalize()
        data = np.copy(kernel.array)

        kernel.normalize()
        assert_allclose(data, kernel.array)

        kernel.normalize()
        assert_allclose(data, kernel.array)

    def test_kernel_normalization_mode(self):
        """
        Test that an error is raised if mode is invalid.
        """
        with pytest.raises(ValueError):
            kernel = CustomKernel(np.ones(3))
            kernel.normalize(mode='invalid')

    def test_kernel1d_int_size(self):
        """
        Test that an error is raised if ``Kernel1D`` ``x_size`` is not
        an integer.
        """
        with pytest.raises(TypeError):
            Gaussian1DKernel(3, x_size=1.2)

    def test_kernel2d_int_xsize(self):
        """
        Test that an error is raised if ``Kernel2D`` ``x_size`` is not
        an integer.
        """
        with pytest.raises(TypeError):
            Gaussian2DKernel(3, x_size=1.2)

    def test_kernel2d_int_ysize(self):
        """
        Test that an error is raised if ``Kernel2D`` ``y_size`` is not
        an integer.
        """
        with pytest.raises(TypeError):
            Gaussian2DKernel(3, x_size=5, y_size=1.2)

    def test_kernel1d_initialization(self):
        """
        Test that an error is raised if an array or model is not
        specified for ``Kernel1D``.
        """
        with pytest.raises(TypeError):
            Kernel1D()

    def test_kernel2d_initialization(self):
        """
        Test that an error is raised if an array or model is not
        specified for ``Kernel2D``.
        """
        with pytest.raises(TypeError):
            Kernel2D()