color_map_2d.py

import numpy as np
import cv2
import scipy.spatial


def get_color_for_point(point_coords, list_of_point_centers, list_of_colors):
    """
    get_color_for_point() computes an RGB color value for a point in the
    (-1, 1) 2D plane, based on a set of RGB color values, defined on particular
    points of the same 2D plane.
    :param point_coords: coordinates for the point for which we want to
    calculate its color
    :param list_of_point_centers: list of point coodinates
    [[x1, y1], ..., [xN, yMN] of the of the aforementioned colors
    :param list_of_colors:  list of RGB color values [[R1, G1, B1], ...,
    [RN, GN, BN]] for the N points in the 2D plane (see prev atribute)
    :return: interpolated RGB color for the input point (1st arg)
    """
    color = np.array([0.0, 0.0, 0.0])
    # get distances of the "query" point from all other points
    distances = scipy.spatial.distance.cdist([point_coords],
                                             list_of_point_centers)[0]
    
    # get weights and compute new RGB value as weighted sum:
    weights = 1 / (distances + 0.1)
    for ic, c in enumerate(list_of_colors):
        color += (np.array(c) * weights[ic])
    color /= (np.sum(weights))
    sum_color = np.sum(color)
    required_sum_color = 600.0
    if color.max() * (required_sum_color/sum_color) <= 255:
        color *= (required_sum_color/sum_color)
    else:
        color *= (255/(color.max()))
    return color


def create_2d_color_map(list_of_points, list_of_colors, height, width):
    """
    create_2d_color_map() creates a colormap by interpolating RGB color values,
    given a list of colors to be defined on particular points of the 2D
    plane.
    :param list_of_points: list of point coodinates
    [[x1, y1], ..., [xN, yMN] of the of the aforementioned colors
    :param list_of_colors: list of RGB color values [[R1, G1, B1], ...,
    [RN, GN, BN]] for the N points in the 2D plane (see prev atribute)
    :param height: output image height
    :param width:  output image weight
    :return: estimated color image
    """
    rgb = np.zeros((height, width, 3)).astype("uint8")
    c_x = int(width / 2)
    c_y = int(height / 2)
    step = 5
    win_size = int((step-1) / 2)
    for i in range(len(list_of_points)):
        rgb[c_y - int(list_of_points[i][1] * height / 2),
            c_x + int(list_of_points[i][0] * width / 2)] = list_of_colors[i]
    for y in range(win_size, height - win_size, step):
        for x in range(win_size, width - win_size, step):
            x_real = (x - width / 2) / (width / 2)
            y_real = (height / 2 - y ) / (height / 2)
            color = get_color_for_point([x_real, y_real], list_of_points,
                                        list_of_colors)
            rgb[y - win_size - 1 : y + win_size + 1,
                x - win_size - 1 : x + win_size + 1] = color
    bgr = cv2.cvtColor(rgb, cv2.COLOR_BGR2RGB)
    return bgr


if __name__ == "__main__":
    colors = {"coral": [255,127,80],
              "pink": [255, 192, 203],
              "orange": [255, 165, 0],
              "blue": [0, 0, 205],
              "green": [0, 205, 0],
              "red": [205, 0, 0],
              "yellow": [204, 204, 0]}
    angry_pos = [-0.8, 0.5]
    fear_pos = [-0.3, 0.8]
    happy_pos = [0.6, 0.6]
    calm_pos = [0.4, -0.5]
    sad_pos = [-0.6, -0.4]
    bgr = create_2d_color_map([angry_pos, fear_pos, happy_pos,
                               calm_pos, sad_pos],
                              [colors["red"],  colors["yellow"],
                               colors["orange"], colors["green"],
                               colors["blue"]], 200, 200)
    cv2.imshow('Signal', bgr)
    ch = cv2.waitKey(10000)