Skip to content

Color constancy

ColorConstancy ¤

Bases: Subfunction_Base

A Subfunction class which fixes the problem of Color Constancy in an image.

Warning

Can only be applied on RGB images.

"The paper Improving dermoscopy image classification using color constancy shows that using a color compensation technique to reduce the influence of the acquisition setup on the color features extracted from the images provides a improvement on the performance for skin cancer classification."

Cite

Description from: https://www.kaggle.com/apacheco/shades-of-gray-color-constancy

Reference - Implementation

https://github.com/nickshawn/Shades_of_Gray-color_constancy_transformation

Reference - Publication

Catarina Barata; M. Emre Celebi; Jorge S. Marques. 2014. Improving Dermoscopy Image Classification Using Color Constancy.
https://ieeexplore.ieee.org/abstract/document/6866131

Source code in aucmedi/data_processing/subfunctions/color_constancy.py 
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
66
67
68
69
70
71
72
73
74
75
76
77
78
79
80
81
82
83
class ColorConstancy(Subfunction_Base):
    """  A Subfunction class which fixes the problem of Color Constancy in an image.

    ???+ warning
        Can only be applied on RGB images.

    "The paper Improving dermoscopy image classification using color constancy shows
    that using a color compensation technique to reduce the influence of the acquisition
    setup on the color features extracted from the images provides a improvement on the
    performance for skin cancer classification."

    ???+ cite
        Description from: https://www.kaggle.com/apacheco/shades-of-gray-color-constancy

    ??? abstract "Reference - Implementation"
        https://github.com/nickshawn/Shades_of_Gray-color_constancy_transformation

    ??? abstract "Reference - Publication"
        Catarina Barata; M. Emre Celebi; Jorge S. Marques. 2014.
        Improving Dermoscopy Image Classification Using Color Constancy.
        <br>
        https://ieeexplore.ieee.org/abstract/document/6866131
    """
    #---------------------------------------------#
    #                Initialization               #
    #---------------------------------------------#
    def __init__(self, power=6):
        """ Initialization function for creating a ColorConstancy Subfunction which can be passed to a
            [DataGenerator][aucmedi.data_processing.data_generator.DataGenerator].

        Args:
            power (int):            Exponent for the image.
        """
        self.power = power

    #---------------------------------------------#
    #                Transformation               #
    #---------------------------------------------#
    def transform(self, image):
        # Verify if image is RGB
        if image.shape[-1] != 3:
            raise ValueError("Image have to be RGB for Color Constancy application!",
                             "Last axis of image is not 3 (RGB):", image.shape)
        # Apply color constancy filtering (Shades of Gray)
        img = image.astype('float32')
        img_power = np.power(img, self.power)
        axes = tuple(range(len(image.shape[:-1])))
        rgb_vec = np.power(np.mean(img_power, axes), 1/self.power)
        rgb_norm = np.sqrt(np.sum(np.power(rgb_vec, 2.0)))
        rgb_vec = rgb_vec / rgb_norm
        rgb_vec = 1 / (rgb_vec * np.sqrt(3))
        img_filtered = np.multiply(img, rgb_vec)
        # Return filtered image
        return img_filtered

__init__(power=6) ¤

Initialization function for creating a ColorConstancy Subfunction which can be passed to a DataGenerator.

Parameters:

Name Type Description Default
power int

Exponent for the image.

 6
Source code in aucmedi/data_processing/subfunctions/color_constancy.py 
56
57
58
59
60
61
62
63
def __init__(self, power=6):
    """ Initialization function for creating a ColorConstancy Subfunction which can be passed to a
        [DataGenerator][aucmedi.data_processing.data_generator.DataGenerator].

    Args:
        power (int):            Exponent for the image.
    """
    self.power = power