current community

your communities

more stack exchange communities

Stack Exchange
tour help
careers 2.0
Take the 2-minute tour ×
Stack Overflow is a question and answer site for professional and enthusiast programmers. It's 100% free, no registration required.
up vote 0 down vote favorite
1

I have a method in my project in which I verify if a pixel has the desired reliability (in terms of its classification as edge or not) and I plot the pixels in the following scheme:

White -> pixel doesn't have the required reliability
Blue -> pixel has the required reliability and it was classified as not edge
Red -> pixel has the required reliability and it was classified as an edge

This is my code:

def generate_data_reliability(classification_mean, data_uncertainty, x_axis_label, y_axis_label, plot_title,
                                  file_path, reliability):
        """
        :classification_mean : given a set of images, how was the mean classification for each pixel
        :param data_uncertainty : the uncertainty about the classification
        :param x_axis_label : the x axis label of the data
        :param y_axis_label : the y axis label of the data
        :param plot_title : the title of the data
        :param file_path : the name of the file
        """
        plt.figure()
        # 0 -> certainty
        # 1 -> uncertainty
        r = 0
        b = 0
        w = 0
        has_reliability = numpy.zeros((data_uncertainty.rows, data_uncertainty.cols), float)
        for x, y in product(range(data_uncertainty.rows), range(data_uncertainty.cols)):
            # I the uncertainty is > then the required reliability, doesn't show it
            if data_uncertainty.data[x][y] > (1.0 - reliability):
                has_reliability[x][y] = 0.5
                w += 1
            else:
                has_reliability[x][y] = classification_mean.data[x][y]
                if has_reliability[x][y] == 1.0:
                    r += 1
                if has_reliability[x][y] == 0.0:
                    b += 1

        print reliability, w+r+b, w, r, b

        plt.title(plot_title)
        plt.imshow(has_reliability, extent=[0, classification_mean.cols, classification_mean.rows, 0], cmap='bwr')
        plt.xlabel(x_axis_label)
        plt.ylabel(y_axis_label)
        plt.savefig(file_path + '.png')
        plt.close()

And this is the print that I got:

>>>> Prewitt
0.8 95100 10329 0 84771
0.9 95100 12380 0 82720
0.99 95100 18577 0 76523

As can be seen, as the required reliability get higher, less pixels have this reliability (more of then will be plot white and none of them are red).

But this is the plots that I get:

enter image description here

enter image description here

enter image description here

I don't know why, if I have less pixels with the desired reliability, I don't get more white pixels, but these red ones. I'm not changing my objects, to mess with them. Oo

I'm stucked in this problem at about 3 hours with no clue about what is wrong.

EDIT:

In this cmap 0 is blue, 0.5 is white and 1 is red, isn't it? I'm pretty sure that the problem is 'cause I am using a diverging color map and sometimes and don't have a central value. E.g., in the situation that I posted here, I don't have red values, so my values vary between 0.5 and 1. Then, matplotlib automatically set my min value to be red and my max value to be blue. But how could I do that? I choose this 'cause would like to represent colors in the scheme: 0=blue, 0.5=white and 1=red (My values will always be 0, 0.5 or 1).

Any help would be very, very much appreciated.

Thank you in advance.

share|improve this question
3  
can you post a minimilistic working example (including sample image to run through your code), so that the plots you have made can be reproduced? –  three_pineapples yesterday
    
Hi, @three_pineapples. have a project that running in parallel with 9 threads takes 1 hour (to generate the input images). It's kind of complicate to provide the code. :S But I discover what is the problem (see EDIT), I just don't know how to solve it yet. :S –  pceccon 12 hours ago
add comment

1 Answer

up vote 0 down vote

Well, I could achieve what I want using a custom colormap. This is the code:

@staticmethod
    def generate_data_reliability(classification_mean, data_uncertainty, x_axis_label, y_axis_label, plot_title,
                                  file_path, reliability):
    """
    :param data_uncertainty : the uncertainty about the data
    :param x_axis_label : the x axis label of the data
    :param y_axis_label : the y axis label of the data
    :param plot_title : the title of the data
    :param file_path : the name of the file
    """
    color_map = mpl.colors.ListedColormap(['blue', 'white', 'red'])
    # From 0 to 0.24 -> blue
    # From 0.25 to 0.4 -> white
    # From 0.5 to 1.0 -> red
    bounds = [0.0, 0.25, 0.5, 1.0]
    norm = mpl.colors.BoundaryNorm(bounds, color_map.N)

    plt.figure()
    # 0 -> certainty
    # 1 -> uncertainty
    r = 0
    b = 0
    w = 0
    has_reliability = numpy.zeros((data_uncertainty.rows, data_uncertainty.cols), float)
    for x, y in product(range(data_uncertainty.rows), range(data_uncertainty.cols)):
        # I the uncertainty is > then the required reliability, doesn't show it
        if data_uncertainty.data[x][y] > (1.0 - reliability):
            has_reliability[x][y] = 0.4
        else:
            has_reliability[x][y] = classification_mean.data[x][y]

    plt.title(plot_title)
    plt.imshow(has_reliability, extent=[0, classification_mean.cols, classification_mean.rows, 0],
               interpolation='nearest', cmap=color_map, norm=norm)
    plt.xlabel(x_axis_label)
    plt.ylabel(y_axis_label)
    plt.savefig(file_path + '.png')
    plt.close()
share|improve this answer
add comment

Your Answer

 
discard

By posting your answer, you agree to the privacy policy and terms of service.

Not the answer you're looking for? Browse other questions tagged or ask your own question.