diff --git a/deepdespeckling/merlin/merlin_denoiser.py b/deepdespeckling/merlin/merlin_denoiser.py
index 2700603..92ae78e 100644
--- a/deepdespeckling/merlin/merlin_denoiser.py
+++ b/deepdespeckling/merlin/merlin_denoiser.py
@@ -1,6 +1,7 @@
 from glob import glob
 import logging
 from pathlib import Path
+from typing import Tuple
 import torch
 import os
 import numpy as np
@@ -8,7 +9,7 @@
 
 from deepdespeckling.denoiser import Denoiser
 from deepdespeckling.model import Model
-from deepdespeckling.utils.constants import M, m
+from deepdespeckling.utils.constants import M, m, M_s, m_s
 from deepdespeckling.utils.utils import (denormalize_sar_image, load_sar_image, save_image_to_npy_and_png,
                                          symetrise_real_and_imaginary_parts, create_empty_folder_in_directory)
 
@@ -28,6 +29,7 @@ def __init__(self, model_name, symetrise, **params):
         super().__init__(**params)
         self.model_name = model_name
         self.symetrise = symetrise
+        self.M, self.m = self.init_min_max_norm_values()
         self.weights_path = self.init_model_weights_path()
 
     def init_model_weights_path(self) -> str:
@@ -51,6 +53,16 @@ def init_model_weights_path(self) -> str:
 
         return model_weights_path
 
+    def init_min_max_norm_values(self) -> Tuple[float, float]:
+        """Initialize min and max values for normalization depending on model name
+
+        Returns:
+            Tuple[float, float]: Max and min values for SAR image normalization
+        """
+        if self.model_name == "Sentinel-TOPS":
+            return M_s, m_s
+        return M, m
+
     def load_model(self, patch_size: int) -> Model:
         """Load model with given weights 
 
@@ -213,9 +225,9 @@ def denoise_image(self, noisy_image: np.array, patch_size: int, stride_size: int
                     imag_to_denoise, device=self.device, dtype=torch.float32)
 
                 real_to_denoise = (torch.log(torch.square(
-                    real_to_denoise)+1e-3)-2*m)/(2*(M-m))
+                    real_to_denoise)+1e-3)-2*self.m)/(2*(self.M-self.m))
                 imag_to_denoise = (torch.log(torch.square(
-                    imag_to_denoise)+1e-3)-2*m)/(2*(M-m))
+                    imag_to_denoise)+1e-3)-2*self.m)/(2*(self.M-self.m))
 
                 denoised_image_real_part = self.denoise_image_kernel(
                     real_to_denoise, denoised_image_real_part, x, y, patch_size, model)
diff --git a/deepdespeckling/utils/constants.py b/deepdespeckling/utils/constants.py
index 0adb93c..f97f468 100644
--- a/deepdespeckling/utils/constants.py
+++ b/deepdespeckling/utils/constants.py
@@ -1,7 +1,13 @@
+import numpy as np
+
 # Global maximum and minimum values obtained empirically and used to normalize SAR images
 M = 10.089038980848645
 m = -1.429329123112601
 
+# Sentinel-TOPS global maximum and minimum constants used to normalize SAR images
+M_s = 12.385
+m_s = -np.log(0.001)
+
 # Must be a power of 2 lower than min(height, width) of the image to despeckle
 # Default to 256 as the trained spotlight and stripmap models stored in merlin/saved_model
 PATCH_SIZE = 256
diff --git a/setup.py b/setup.py
index db896e8..2c5aea5 100644
--- a/setup.py
+++ b/setup.py
@@ -40,6 +40,6 @@
     name='deepdespeckling',
     packages=find_packages(include=['deepdespeckling', 'deepdespeckling.*']),
     url='https://github.com/hi-paris/deepdespeckling',
-    version='0.6',
+    version='0.7',
     zip_safe=False,
 )