Merge pull request #53 from remotesensinginfo/sort_drone_images

Added function to sort a set of drone images

doc/python/source/rsgislib_tools_imagetools.rst

@@ -5,3 +5,11 @@ Animate Images
 -----------------
 .. autofunction:: rsgislib.tools.imagetools.animate_img_set
 
+EXIF Metadata
+-----------------
+.. autofunction:: rsgislib.tools.imagetools.list_exif_info
+
+Sort Images
+-----------------
+.. autofunction:: rsgislib.tools.imagetools.split_photos_by_time
+

python/rsgislib/tools/imagetools.py

@@ -11,6 +11,95 @@
 import rsgislib.tools.utils
 
 
+class ImageTimeCluster(object):
+    """
+    A class which supports the split_photos_by_time function.
+    """
+
+    def __init__(self, cluster_id=None, init_time=None):
+        """
+
+        :param cluster_id: unique id for the cluster
+        :param init_time: is the initial time of the first image
+
+        """
+        self.cluster_id = cluster_id
+        self.start_time = init_time
+        self.end_time = init_time
+        self.images = list()
+
+    def within_cluster(self, date_time_obj, time_split_secs=60):
+        date_time_in_cluster = False
+        start_diff = self.start_time - date_time_obj
+        if abs(start_diff.total_seconds()) < time_split_secs:
+            date_time_in_cluster = True
+
+        if not date_time_in_cluster:
+            end_diff = self.end_time - date_time_obj
+            if abs(end_diff.total_seconds()) < time_split_secs:
+                date_time_in_cluster = True
+        return date_time_in_cluster
+
+    def add_image_to_cluster(self, image_file, date_time_obj):
+        self.images.append(image_file)
+        if date_time_obj < self.start_time:
+            self.start_time = date_time_obj
+        elif date_time_obj > self.end_time:
+            self.end_time = date_time_obj
+
+    def does_cluster_overlap(self, cluster_obj, time_split_secs=60):
+        date_time_in_cluster = False
+        time_diff = self.start_time - cluster_obj.start_time
+        if abs(time_diff.total_seconds()) < time_split_secs:
+            date_time_in_cluster = True
+
+        if not date_time_in_cluster:
+            time_diff = self.end_time - cluster_obj.end_time
+            if abs(time_diff.total_seconds()) < time_split_secs:
+                date_time_in_cluster = True
+
+        if not date_time_in_cluster:
+            time_diff = self.start_time - cluster_obj.end_time
+            if abs(time_diff.total_seconds()) < time_split_secs:
+                date_time_in_cluster = True
+
+        if not date_time_in_cluster:
+            time_diff = self.end_time - cluster_obj.start_time
+            if abs(time_diff.total_seconds()) < time_split_secs:
+                date_time_in_cluster = True
+
+        if not date_time_in_cluster:
+            if (cluster_obj.start_time > self.start_time) and (
+                cluster_obj.start_time < self.end_time
+            ):
+                date_time_in_cluster = True
+
+        if not date_time_in_cluster:
+            if (cluster_obj.end_time > self.start_time) and (
+                cluster_obj.end_time < self.end_time
+            ):
+                date_time_in_cluster = True
+
+        return date_time_in_cluster
+
+    def merge_cluster(self, cluster_obj):
+        self.images = self.images + cluster_obj.images
+        if cluster_obj.start_time < self.start_time:
+            self.start_time = cluster_obj.start_time
+        elif cluster_obj.end_time > self.end_time:
+            self.end_time = cluster_obj.end_time
+
+    def __str__(self):
+        return "{}: {} - {} = {}".format(
+            self.cluster_id, self.start_time, self.end_time, len(self.images)
+        )
+
+    def __repr__(self):
+        return "{}: {} - {} = {}".format(
+            self.cluster_id, self.start_time, self.end_time, len(self.images)
+        )
+
+
 def animate_img_set(
     input_imgs: List[str],
     output_img: str,
@@ -78,3 +167,133 @@ def animate_img_set(
 
     if resize:
         shutil.rmtree(lcl_tmp_dir)
+
+
+def list_exif_info(input_img):
+    """
+    A function which lists the exif information from an input image
+    using the PIL module.
+
+    :param input_img: the path to the input image
+
+    """
+    from PIL import Image
+    from PIL.ExifTags import TAGS
+
+    pil_image = Image.open(input_img)
+    pil_exif_data = pil_image.getexif()
+
+    for tag_id in pil_exif_data:
+        # get the tag name, instead of human unreadable tag id
+        tag = TAGS.get(tag_id, tag_id)
+        data = pil_exif_data.get(tag_id)
+        # decode bytes
+        if isinstance(data, bytes):
+            data = data.decode()
+        print(f"{tag:25}: {data}")
+    pil_image.close()
+
+
+def split_photos_by_time(
+    input_imgs,
+    output_base_dir,
+    time_split_secs=60,
+    date_time_tag="DateTime",
+    date_time_format="%Y:%m:%d %H:%M:%S",
+    out_date_time_format="%Y%m%d_%H%M%S",
+):
+    """
+    A function which splits a set of images (e.g., from drone surveys) by time.
+    Where photos taken within the specified time window will be clustered together
+    (default: 60 seconds). The photos are then copied into a new directory based on
+    the start time of the cluster. Note, photos are copied and not moved from their
+    current location.
+
+    :param input_imgs: a list of input images (e.g., glob.glob("*.jpg"))
+    :param output_base_dir: the output base directory where the output directories
+                            will be created.
+    :param time_split_secs: The time difference in seconds for a new cluster to be
+                            created.
+    :param date_time_tag: the EXIF tag for the date / time of the acquisition.
+                          Default is 'DataTime' but this might be different for
+                          different cameras and therefore you might need to use the
+                          list_exif_info function to find the correct field.
+    :param date_time_format: The python datetime.strptime format string for the
+                             datetime provided in the EXIF metadata.
+                             Default: %Y:%m:%d %H:%M:%S
+    :param out_date_time_format: The python datetime.strftime format string for the
+                                 used to create the output directory for the files.
+                                 Note, this must be unique. Default: %Y%m%d_%H%M%S
+
+    """
+
+    from PIL import Image
+    from PIL.ExifTags import TAGS
+    import datetime
+    import tqdm
+
+    img_dt_lut = dict()
+    for img_file in input_imgs:
+        pil_image = Image.open(img_file)
+        pil_exif_data = pil_image.getexif()
+
+        for tag_id in pil_exif_data:
+            # get the tag name, instead of human unreadable tag id
+            tag = TAGS.get(tag_id, tag_id)
+            data = pil_exif_data.get(tag_id)
+            # decode bytes
+            if isinstance(data, bytes):
+                data = data.decode()
+            if tag == date_time_tag:
+                img_dt_lut[img_file] = data
+
+    img_clusters = list()
+    next_cluster_id = 0
+    first = True
+    for img_file in img_dt_lut:
+        date_time_obj = datetime.datetime.strptime(
+            img_dt_lut[img_file], date_time_format
+        )
+
+        if first:
+            img_cluster_obj = ImageTimeCluster(next_cluster_id, date_time_obj)
+            img_cluster_obj.images.append(img_file)
+            img_clusters.append(img_cluster_obj)
+            next_cluster_id += 1
+            first = False
+        else:
+            for img_cluster in img_clusters:
+                if img_cluster.within_cluster(date_time_obj, time_split_secs):
+                    img_cluster.add_image_to_cluster(img_file, date_time_obj)
+                    break
+                else:
+                    img_cluster_obj = ImageTimeCluster(next_cluster_id, date_time_obj)
+                    img_cluster_obj.images.append(img_file)
+                    img_clusters.append(img_cluster_obj)
+                    next_cluster_id += 1
+                    break
+
+        merged_cluster = False
+        for img_cluster_out in img_clusters:
+            for img_cluster_in in img_clusters:
+                if img_cluster_out.cluster_id != img_cluster_in.cluster_id:
+                    cluster_overlap = img_cluster_out.does_cluster_overlap(
+                        img_cluster_in
+                    )
+                    if cluster_overlap:
+                        merged_cluster = True
+                        img_cluster_out.merge_cluster(img_cluster_in)
+                        img_clusters.remove(img_cluster_in)
+                        break
+            if merged_cluster:
+                break
+
+    for img_cluster in img_clusters:
+        print("Copying Files for: {}".format(img_cluster))
+        out_dir_name = img_cluster.start_time.strftime(out_date_time_format)
+        out_dir_path = os.path.join(output_base_dir, out_dir_name)
+        print("Output Directory: {}".format(out_dir_path))
+        if not os.path.exists(out_dir_path):
+            os.mkdir(out_dir_path)
+        for img in tqdm.tqdm(img_cluster.images):
+            shutil.copy(img, out_dir_path)

python_tests/test_tools_ftptools.py

@@ -22,7 +22,7 @@
 ftp_user_ref = "dlpuser"
 ftp_pass_ref = "rNrKYTX9g7z3RgJRmxWuGHbeu"
 
-
+@pytest.mark.skipif(True, reason="Skip test as timeout can be a problem for ci.")
 def test_upload_ftp_file():
     import rsgislib.tools.ftptools
 
@@ -63,10 +63,11 @@ def test_get_ftp_file_listings():
     assert len(non_dirs_lst) > 0
 
 
-@pytest.mark.skipif(
-    True,
-    reason="pysondb dependency not available and test currently skipped as timeout can be a problem for ci.",
-)
+#@pytest.mark.skipif(
+#    True,
+#    reason="pysondb dependency not available and test currently skipped as timeout can be a problem for ci.",
+#)
+@pytest.mark.skipif(True, reason="Skip test as timeout can be a problem for ci.")
 def test_create_file_listings_db(tmp_path):
     import rsgislib.tools.ftptools
 
@@ -93,7 +94,7 @@ def test_create_file_listings_db(tmp_path):
 
     assert os.path.exists(out_db_file)
 
-
+@pytest.mark.skipif(True, reason="Skip test as timeout can be a problem for ci.")
 def test_download_ftp_file(tmp_path):
     import rsgislib.tools.ftptools
 
@@ -121,7 +122,8 @@ def test_download_ftp_file(tmp_path):
     assert os.path.exists(out_file)
 
 
-@pytest.mark.skipif(PYCURL_DB_NOT_AVAIL, reason="pycurl dependency not available")
+#@pytest.mark.skipif(PYCURL_DB_NOT_AVAIL, reason="pycurl dependency not available")
+@pytest.mark.skipif(True, reason="Skip test as timeout can be a problem for ci.")
 def test_download_curl_ftp_file(tmp_path):
     import rsgislib.tools.ftptools