大津の方法の2値化を実装

__init__.py

@@ -1,11 +1,13 @@
 from .nodes.grayscale_image import GrayscaleImage
 from .nodes.brightness_transparency import BrightnessTransparency
 from .nodes.binarize_image import BinarizeImage
+from .nodes.binarize_image_using_otsu import BinarizeImageUsingOtsu
 from .nodes.antialiasing_image import AntialiasingImage
 
 NODE_CLASS_MAPPINGS = {
     "BrightnessTransparency": BrightnessTransparency,
     "BinarizeImage": BinarizeImage,
+    "BinarizeImageUsingOtsu": BinarizeImageUsingOtsu,
     "GrayscaleImage": GrayscaleImage,
     "AntialiasingImage": AntialiasingImage,
 }
@@ -13,6 +15,7 @@ NODE_CLASS_MAPPINGS = {
 NODE_DISPLAY_NAME_MAPPINGS = {
     "BrightnessTransparency": "BrightnessTransparency",
     "BinarizeImage": "BinarizeImage",
+    "BinarizeImageUsingOtsu": "BinarizeImageUsingOtsu",
     "GrayscaleImage": "GrayscaleImage",
     "AntialiasingImage": "AntialiasingImage",
 }

nodes/binarize_image_using_otsu.py

@@ -0,0 +1,67 @@
+import numpy as np
+import torch
+import cv2
+from ..utils import convert_to_grayscale
+
+
+class BinarizeImageUsingOtsu:
+    @classmethod
+    def INPUT_TYPES(s):
+        return {
+            "required": {
+                "images": ("IMAGE", {"tooltip": "Images to binarize."}),
+            }
+        }
+
+    RETURN_TYPES = ("IMAGE",)
+    FUNCTION = "run"
+    CATEGORY = "ComfyUI-Image-Toolkit"
+
+    def run(self, images):
+        is_tensor = isinstance(images, torch.Tensor)
+
+        if is_tensor:
+            if images.device.type != "cpu":
+                images = images.cpu()
+            images_np = images.numpy()
+        else:
+            images_np = images
+
+        # Process each image in the batch
+        result_images = []
+        for image in images_np:
+            # Convert to grayscale if needed
+            if image.shape[-1] == 3:
+                # Convert to grayscale using our utility function
+                gray = convert_to_grayscale(image)[
+                    :, :, 0
+                ]  # Take only one channel
+            else:
+                gray = image
+
+            # Convert to uint8 for OpenCV
+            gray_uint8 = (gray * 255).astype(np.uint8)
+
+            # Apply Otsu's thresholding
+            _, binary = cv2.threshold(
+                gray_uint8, 0, 255, cv2.THRESH_BINARY + cv2.THRESH_OTSU
+            )
+
+            # Convert back to float32 and normalize
+            binary_float = binary.astype(np.float32) / 255.0
+
+            # Expand to 3 channels for consistency
+            binary_float = np.stack([binary_float] * 3, axis=-1)
+            result_images.append(binary_float)
+
+        # Stack the results back into a batch
+        result_np = np.stack(result_images)
+
+        # Convert back to tensor if input was tensor
+        if is_tensor:
+            result_tensor = torch.from_numpy(result_np)
+            if images.device.type != "cpu":
+                result_tensor = result_tensor.to(images.device)
+            return (result_tensor,)
+
+        return (result_np,)