Code to convert polar to cartesian radar

[Madhusakth]

Hi,

Would it be possible to share the code you used for converting the radar data from polar to cartesian?

Thanks!

[Ianpengg]

Hi~
You can take this code from Oxford radar-robotcar-dataset for reference
I've already modified a little bit to fit the polar data provided by Radiate.

def from_polar_to_cart(raw_example_data, cart_resolution: float, cart_pixel_width: int, interpolate_crossover=True):

    radar_resolution = np.array([0.0432], np.float32)
    azimuths = np.arange(400) * (math.pi/180)*0.9
    raw_example_data = raw_example_data.T
    fft_data = raw_example_data[:, :].astype(np.float32)[:, :, np.newaxis] / 255.
    if (cart_pixel_width % 2) == 0:
        cart_min_range = (cart_pixel_width / 2 - 0.5) * cart_resolution
    else:
        cart_min_range = cart_pixel_width // 2 * cart_resolution
    coords = np.linspace(-cart_min_range, cart_min_range, cart_pixel_width, dtype=np.float32)
    Y, X = np.meshgrid(coords, -coords)
    sample_range = np.sqrt(Y * Y + X * X)
    sample_angle = np.arctan2(Y, X)
    sample_angle += (sample_angle < 0).astype(np.float32) * 2. * np.pi

    # Interpolate Radar Data Coordinates
    azimuth_step = azimuths[1] - azimuths[0]

    sample_u = (sample_range - radar_resolution / 2) / radar_resolution
    sample_v = (sample_angle - azimuths[0]) / azimuth_step

    # We clip the sample points to the minimum sensor reading range so that we
    # do not have undefined results in the centre of the image. In practice
    # this region is simply undefined.
    sample_u[sample_u < 0] = 0

    if interpolate_crossover:
        fft_data = np.concatenate((fft_data[-1:], fft_data, fft_data[:1]), 0)
        sample_v = sample_v + 1

    polar_to_cart_warp = np.stack((sample_u, sample_v), -1)
    cart_img = np.expand_dims(cv2.remap(fft_data, polar_to_cart_warp, None, cv2.INTER_LINEAR), -1)
    
    return cart_img

[yangsc4063]

However, after using the above method, I found that there was still a difference between the images provided by the dataset and the images I transformed. You can zoom in and compare carefully, or use the following line of code

# compare the interpolated BEV image with the Cartesian radar image
assert np.allclose(bev_img, cart_img, atol=1e-1)

I am also very curious about code that convert polar to cartesian. Looking forward to your reply!