2d np 数组到 3d np 数组

Question

我有一个 2D np.array，我必须将其可视化。我已经有了一个颜色图函数，所以我要做的最后一件事就是用另外 3 个值替换数组的每个值，这些值代表 RGB。

颜色图功能：

def colormapHLS(a: float):
    h, l, s = 0, 0, 0
    if a < 0:
        h = 2 / 3
        l = 0
        s = 1
    elif 0 < a <= 1 / 3:
        h = 2 / 3
        l = 0.5 * (a * 3)
        s = 1
    elif 1 / 3 < a <= 2 / 3:
        h = 2 / 3 - ((2 / 3) * ((a - 1 / 3) * 3))
        l = 0.5
        s = 1
    elif 2 / 3 < a <= 1:
        h = 0
        l = 0.5 + 0.5 * ((a - 2 / 3) * 3)
        s = 1
    elif a > 1:
        h = 0
        l = 1
        s = 1
    return h, l, s

和主要代码：

# ...
trans_func = lambda x: colorsys.hls_to_rgb(*colormapHLS(x))

draw_buff -= numpy.min(draw_buff)
draw_buff /= numpy.max(draw_buff)
print(draw_buff.shape)
colored_draw_buff = numpy.zeros((*draw_buff.shape, 3))
print(colored_draw_buff.shape)
for y in range(draw_buff.shape[0]):
    for x in range(draw_buff.shape[1]):
        cur = trans_func(draw_buff[y][x])
        for z in range(3):
            colored_draw_buff[y][x][z] = cur[z]

colored_draw_buff *= 255
# ...

绘制buff形状为(x,y)

所以，它确实有效，但是太慢了。我认为有一个更快的方法

Answer 1

        self.draw_buff = self.draw_buff.reshape(x*y)

        self.draw_buff -= numpy.min(self.draw_buff)
        self.draw_buff /= numpy.max(self.draw_buff)

        self.draw_buff = numpy.array(trans_func(self.draw_buff))
        self.draw_buff *= 255
        self.draw_buff = self.draw_buff.swapaxes(0,1)
        self.draw_buff = self.draw_buff.reshape((y,x,3))

交换轴+重塑正在完成这项工作