普通文本  |  175行  |  7.15 KB

# Copyright 2015 The Android Open Source Project
#
# Licensed under the Apache License, Version 2.0 (the "License");
# you may not use this file except in compliance with the License.
# You may obtain a copy of the License at
#
#      http://www.apache.org/licenses/LICENSE-2.0
#
# Unless required by applicable law or agreed to in writing, software
# distributed under the License is distributed on an "AS IS" BASIS,
# WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
# See the License for the specific language governing permissions and
# limitations under the License.

import its.image
import its.caps
import its.device
import its.objects
import its.target
import matplotlib
import matplotlib.pyplot
import numpy
import os.path
from matplotlib import pylab

NR_MODES = [0, 1, 2, 3, 4]


def main():
    """Test that the android.noiseReduction.mode param is applied when set for
       reprocessing requests.

    Capture reprocessed images with the camera dimly lit. Uses a high analog
    gain to ensure the captured image is noisy.

    Captures three reprocessed images, for NR off, "fast", and "high quality".
    Also captures a reprocessed image with low gain and NR off, and uses the
    variance of this as the baseline.
    """

    NAME = os.path.basename(__file__).split(".")[0]

    NUM_SAMPLES_PER_MODE = 4
    SNR_TOLERANCE = 3 # unit in db

    with its.device.ItsSession() as cam:
        props = cam.get_camera_properties()

        its.caps.skip_unless(its.caps.compute_target_exposure(props) and
                             its.caps.per_frame_control(props) and
                             its.caps.noise_reduction_mode(props, 0) and
                             (its.caps.yuv_reprocess(props) or
                              its.caps.private_reprocess(props)))

        # If reprocessing is supported, ZSL NR mode must be avaiable.
        assert(its.caps.noise_reduction_mode(props, 4))

        reprocess_formats = []
        if (its.caps.yuv_reprocess(props)):
            reprocess_formats.append("yuv")
        if (its.caps.private_reprocess(props)):
            reprocess_formats.append("private")

        for reprocess_format in reprocess_formats:
            # List of variances for R, G, B.
            snrs = [[], [], []]
            nr_modes_reported = []

            # NR mode 0 with low gain
            e, s = its.target.get_target_exposure_combos(cam)["minSensitivity"]
            req = its.objects.manual_capture_request(s, e)
            req["android.noiseReduction.mode"] = 0

            # Test reprocess_format->JPEG reprocessing
            # TODO: Switch to reprocess_format->YUV when YUV reprocessing is
            #       supported.
            size = its.objects.get_available_output_sizes("jpg", props)[0]
            out_surface = {"width":size[0], "height":size[1], "format":"jpg"}
            cap = cam.do_capture(req, out_surface, reprocess_format)
            img = its.image.decompress_jpeg_to_rgb_image(cap["data"])
            its.image.write_image(img, "%s_low_gain_fmt=jpg.jpg" % (NAME))
            tile = its.image.get_image_patch(img, 0.45, 0.45, 0.1, 0.1)
            ref_snr = its.image.compute_image_snrs(tile)
            print "Ref SNRs:", ref_snr

            e, s = its.target.get_target_exposure_combos(cam)["maxSensitivity"]
            for nr_mode in NR_MODES:
                # Skip unavailable modes
                if not its.caps.noise_reduction_mode(props, nr_mode):
                    nr_modes_reported.append(nr_mode)
                    for channel in range(3):
                        snrs[channel].append(0)
                    continue

                rgb_snr_list = []
                # Capture several images to account for per frame noise
                # variations
                for n in range(NUM_SAMPLES_PER_MODE):
                    req = its.objects.manual_capture_request(s, e)
                    req["android.noiseReduction.mode"] = nr_mode
                    cap = cam.do_capture(req, out_surface, reprocess_format)

                    img = its.image.decompress_jpeg_to_rgb_image(cap["data"])
                    if n == 0:
                        its.image.write_image(
                                img,
                                "%s_high_gain_nr=%d_fmt=jpg.jpg"
                                        %(NAME, nr_mode))
                        nr_modes_reported.append(
                                cap["metadata"]["android.noiseReduction.mode"])

                    tile = its.image.get_image_patch(img, 0.45, 0.45, 0.1, 0.1)
                    # Get the variances for R, G, and B channels
                    rgb_snrs = its.image.compute_image_snrs(tile)
                    rgb_snr_list.append(rgb_snrs)

                r_snrs = [rgb[0] for rgb in rgb_snr_list]
                g_snrs = [rgb[1] for rgb in rgb_snr_list]
                b_snrs = [rgb[2] for rgb in rgb_snr_list]
                rgb_snrs = [numpy.mean(r_snrs),
                            numpy.mean(g_snrs),
                            numpy.mean(b_snrs)]
                print "NR mode", nr_mode, "SNRs:"
                print "    R SNR:", rgb_snrs[0],\
                        "Min:", min(r_snrs), "Max:", max(r_snrs)
                print "    G SNR:", rgb_snrs[1],\
                        "Min:", min(g_snrs), "Max:", max(g_snrs)
                print "    B SNR:", rgb_snrs[2],\
                        "Min:", min(b_snrs), "Max:", max(b_snrs)

                for chan in range(3):
                    snrs[chan].append(rgb_snrs[chan])

            # Draw a plot.
            pylab.figure(reprocess_format)
            for channel in range(3):
                pylab.plot(NR_MODES, snrs[channel], "-"+"rgb"[channel]+"o")

            pylab.xlabel("Noise Reduction Mode")
            pylab.ylabel("SNR (dB)")
            pylab.xticks(NR_MODES)
            matplotlib.pyplot.savefig("%s_plot_%s_SNRs.png" %
                                      (NAME, reprocess_format))

            assert nr_modes_reported == NR_MODES

            for j in range(3):
                # Larger is better
                # Verify OFF(0) is not better than FAST(1)
                assert(snrs[j][0] <
                       snrs[j][1] + SNR_TOLERANCE)
                # Verify FAST(1) is not better than HQ(2)
                assert(snrs[j][1] <
                       snrs[j][2] + SNR_TOLERANCE)
                # Verify HQ(2) is better than OFF(0)
                assert(snrs[j][0] < snrs[j][2])
                if its.caps.noise_reduction_mode(props, 3):
                    # Verify OFF(0) is not better than MINIMAL(3)
                    assert(snrs[j][0] <
                           snrs[j][3] + SNR_TOLERANCE)
                    # Verify MINIMAL(3) is not better than HQ(2)
                    assert(snrs[j][3] <
                           snrs[j][2] + SNR_TOLERANCE)
                    # Verify ZSL(4) is close to MINIMAL(3)
                    assert(numpy.isclose(snrs[j][4], snrs[j][3],
                                         atol=SNR_TOLERANCE))
                else:
                    # Verify ZSL(4) is close to OFF(0)
                    assert(numpy.isclose(snrs[j][4], snrs[j][0],
                                         atol=SNR_TOLERANCE))

if __name__ == '__main__':
    main()