|TensorFlow 1 version||View source on GitHub|
Computes SSIM index between img1 and img2.
Compat aliases for migration
See Migration guide for more details.
tf.image.ssim( img1, img2, max_val, filter_size=11, filter_sigma=1.5, k1=0.01, k2=0.03 )
This function is based on the standard SSIM implementation from: Wang, Z., Bovik, A. C., Sheikh, H. R., & Simoncelli, E. P. (2004). Image quality assessment: from error visibility to structural similarity. IEEE transactions on image processing.
- 11x11 Gaussian filter of width 1.5 is used.
- k1 = 0.01, k2 = 0.03 as in the original paper.
The image sizes must be at least 11x11 because of the filter size.
# Read images (of size 255 x 255) from file. im1 = tf.image.decode_image(tf.io.read_file('path/to/im1.png')) im2 = tf.image.decode_image(tf.io.read_file('path/to/im2.png')) tf.shape(im1) # `img1.png` has 3 channels; shape is `(255, 255, 3)` tf.shape(im2) # `img2.png` has 3 channels; shape is `(255, 255, 3)` # Add an outer batch for each image. im1 = tf.expand_dims(im1, axis=0) im2 = tf.expand_dims(im2, axis=0) # Compute SSIM over tf.uint8 Tensors. ssim1 = tf.image.ssim(im1, im2, max_val=255, filter_size=11, filter_sigma=1.5, k1=0.01, k2=0.03) # Compute SSIM over tf.float32 Tensors. im1 = tf.image.convert_image_dtype(im1, tf.float32) im2 = tf.image.convert_image_dtype(im2, tf.float32) ssim2 = tf.image.ssim(im1, im2, max_val=1.0, filter_size=11, filter_sigma=1.5, k1=0.01, k2=0.03) # ssim1 and ssim2 both have type tf.float32 and are almost equal.
First image batch. 4-D Tensor of shape
Second image batch. 4-D Tensor of shape
||The dynamic range of the images (i.e., the difference between the maximum the and minimum allowed values).|
||Default value 11 (size of gaussian filter).|
||Default value 1.5 (width of gaussian filter).|
||Default value 0.01|
||Default value 0.03 (SSIM is less sensitivity to K2 for lower values, so it would be better if we took the values in the range of 0 < K2 < 0.4).|
|A tensor containing an SSIM value for each image in batch. Returned SSIM values are in range (-1, 1], when pixel values are non-negative. Returns a tensor with shape: broadcast(img1.shape[:-3], img2.shape[:-3]).|