improve image hashing

This commit is contained in:
Ross
2023-12-26 17:11:02 +00:00
parent 8c0aaa82e2
commit 714f472da6
14 changed files with 183 additions and 42 deletions
+11 -11
View File
@@ -58,10 +58,10 @@ def upload_dicom(request, files: List[UploadedFile] = File(...)):
ud.save()
uploaded.append((file.name, ud.image_md5_hash))
uploaded.append((file.name, ud.image_blake3_hash))
except DuplicateDicom:
print(ud.check_for_duplicates(ud.image_md5_hash))
duplicate.append((file.name, ud.image_md5_hash))
print(ud.check_for_duplicates(ud.image_blake3_hash))
duplicate.append((file.name, ud.image_blake3_hash))
pass
return {"uploaded": uploaded, "duplicates": duplicate}
@@ -91,7 +91,7 @@ def uncategorised_dicoms(request):
tags = d.get_basic_dicom_tags()
data[tags["SeriesInstanceUID"]].append(tags)
data["image_hash"].append(d.image_md5_hash)
data["image_hash"].append(d.image_blake3_hash)
return data
@@ -137,7 +137,7 @@ def import_dicoms_helper(request, case_id: int | None = None):
image=dicom.image,
series=series,
is_dicom=True,
image_md5_hash=dicom.image_md5_hash,
image_blake3_hash=dicom.image_blake3_hash,
)
series_image.save()
# dicom.image.delete()
@@ -171,9 +171,9 @@ def series_remove_duplicate_images(request, series_id: int):
img_ids = set()
dupes = set()
for series_image in series.images.all():
if series_image.image_md5_hash in img_ids:
if series_image.image_blake3_hash in img_ids:
dupes.add(series_image.id)
img_ids.add(series_image.image_md5_hash)
img_ids.add(series_image.image_blake3_hash)
if dupes:
SeriesImage.objects.filter(id__in=dupes).delete()
@@ -210,7 +210,7 @@ def get_cases_user(request):
def check_image_hash(request, hash: str):
try:
print(hash)
series_image = SeriesImage.objects.get(image_md5_hash=hash)
series_image = SeriesImage.objects.get(image_blake3_hash=hash)
data = {
"status": "success",
"id": series_image.pk,
@@ -234,7 +234,7 @@ def check_images_hashes(request, hashes: List[str]):
for hash in hashes:
try:
print(hash)
series_image = SeriesImage.objects.get(image_md5_hash=hash)
series_image = SeriesImage.objects.get(image_blake3_hash=hash)
data = {
"id": series_image.pk,
"url": series_image.series.get_absolute_url(),
@@ -250,7 +250,7 @@ def check_images_hashes(request, hashes: List[str]):
def generate_image_hash(request, id: int):
series_image = SeriesImage.objects.get(pk=id)
series_image.image_md5_hash = ""
series_image.image_blake3_hash = ""
series_image.save()
@@ -259,7 +259,7 @@ def generate_image_hash(request, id: int):
@router.get("/view_dicom_tags/{hash}", auth=django_auth)
def view_dicom_tags(request, hash: str):
item = SeriesImage.objects.get(image_md5_hash=hash)
item = SeriesImage.objects.get(image_blake3_hash=hash)
return item.get_dicom_json()
@router.get("/series_split_by_dicom_tag/{series_id}/{dicom_tag}", auth=django_auth)
+1 -1
View File
@@ -444,7 +444,7 @@ CaseCollectionCaseFormSet = inlineformset_factory(
SeriesImageFormSet = inlineformset_factory(
Series,
SeriesImage,
exclude=["dicom_tags_ohif", "replaced", "image_md5_hash", "is_dicom"],
exclude=["dicom_tags_ohif", "replaced", "image_md5_hash", "image_blake3_hash", "is_dicom"],
can_delete=True,
extra=0,
max_num=2000,
@@ -0,0 +1,32 @@
# Generated by Django 4.1.4 on 2023-12-26 16:06
from django.db import migrations, models
class Migration(migrations.Migration):
dependencies = [
('atlas', '0034_uncategoriseddicom_basic_dicom_tags_and_more'),
]
operations = [
migrations.RemoveField(
model_name='uncategoriseddicom',
name='image_md5_hash',
),
migrations.AddField(
model_name='seriesimage',
name='image_blake3_hash',
field=models.CharField(blank=True, max_length=32, null=True),
),
migrations.AddField(
model_name='uncategoriseddicom',
name='image_blake3_hash',
field=models.CharField(blank=True, max_length=32, null=True),
),
migrations.AlterField(
model_name='uncategoriseddicom',
name='basic_dicom_tags',
field=models.JSONField(blank=True, null=True),
),
]
@@ -0,0 +1,18 @@
# Generated by Django 4.1.4 on 2023-12-26 16:29
from django.db import migrations, models
class Migration(migrations.Migration):
dependencies = [
('atlas', '0035_remove_uncategoriseddicom_image_md5_hash_and_more'),
]
operations = [
migrations.AlterField(
model_name='uncategoriseddicom',
name='image_blake3_hash',
field=models.CharField(blank=True, max_length=64, null=True),
),
]
@@ -0,0 +1,18 @@
# Generated by Django 4.1.4 on 2023-12-26 16:34
from django.db import migrations, models
class Migration(migrations.Migration):
dependencies = [
('atlas', '0036_alter_uncategoriseddicom_image_blake3_hash'),
]
operations = [
migrations.AlterField(
model_name='seriesimage',
name='image_blake3_hash',
field=models.CharField(blank=True, max_length=64, null=True),
),
]
+9 -10
View File
@@ -892,7 +892,7 @@ class UncategorisedDicom(models.Model):
series_instance_uid = models.CharField(max_length=255, blank=True, null=True)
image_md5_hash = models.CharField(max_length=32, null=True, blank=True)
image_blake3_hash = models.CharField(max_length=64, null=True, blank=True)
created_date = models.DateTimeField(default=timezone.now)
@@ -901,12 +901,13 @@ class UncategorisedDicom(models.Model):
def check_for_duplicates(self, image_hash):
duplicate = None
if obj := UncategorisedDicom.objects.filter(image_md5_hash=image_hash).first():
duplicate = obj
if obj := UncategorisedDicom.objects.filter(image_blake3_hash=image_hash).first():
#duplicate = obj
return obj
print(image_hash)
if obj := SeriesImage.objects.filter(image_md5_hash=image_hash).first():
if obj := SeriesImage.objects.filter(image_blake3_hash=image_hash).first():
duplicate = obj
return duplicate
@@ -920,11 +921,11 @@ class UncategorisedDicom(models.Model):
# if self.check_for_duplicates():
# return DuplicateDicom
image_hash = get_image_dicom_hash(None, dataset=ds)
image_blake3_hash = get_image_dicom_hash(None, dataset=ds, hash_type="blake3", direct_pixel_data=True)
self.image_md5_hash = image_hash
self.image_blake3_hash = image_blake3_hash
duplicate = self.check_for_duplicates(image_hash)
duplicate = self.check_for_duplicates(image_blake3_hash)
self.basic_dicom_tags = self.get_basic_dicom_tags(dataset=ds)
@@ -932,10 +933,8 @@ class UncategorisedDicom(models.Model):
if duplicate != self:
raise DuplicateDicom
# Hack for tests
if image_hash != "1234":
if image_blake3_hash != "1234":
super().save(*args, **kwargs) # Call the "real" save() method.
def get_dicom_info(self):
+1 -1
View File
@@ -100,7 +100,7 @@
[{{image.image.size|filesizeformat}}]
{% endif %}
{{image.image_md5_hash}} ({{image.is_dicom}})
{{image.image_md5_hash}} ({{image.is_dicom}}) {{image.image_blake3_hash}}
<br />
{% comment %} {{image.get_dicom_info|safe}}<br /> {% endcomment %}