changes image quality check

tool_utils/image_metadata.py

@@ -3,14 +3,15 @@ import numpy as np
 import cv2
 from skimage.restoration import estimate_sigma
 import logging
+
 def image_brightness(image,thresh=0.37):
     L,A,B = cv2.split(cv2.cvtColor(image,cv2.COLOR_BGR2LAB))
     norm_L = L/np.max(L)
     L_mean = np.mean(norm_L)
     if L_mean > thresh:
-        return "image is Bright enough  "
+        return "Image is bright enough to process object detection and segmentation task"
     else:
-        return "image is not bright enough "
+        return "image is not bright enough to process object detection and segmentation task "
 
 def variance_of_laplacian(img,threshould=250):
     # compute the Laplacian of the image and then return the focus
@@ -20,20 +21,20 @@ def variance_of_laplacian(img,threshould=250):
     laplacian_value = cv2.Laplacian(gray, cv2.CV_64F).var()
     logging.info(laplacian_value)
     if laplacian_value <= threshould:
-        return " Image is very blurry"
+        return "There is Blur/Gaussian noise in the image"
     elif laplacian_value <= 3*threshould:
-        return " Image is visible but have some regions out of foucs."
+        return " There is Blur/Gaussian noise in few regions of the image."
     elif laplacian_value >= 3*threshould:
-        return "Image is Very Sharp." 
+        return "Image has high sharpness , no need to process futher." 
 
 def get_signal_to_noise_ratio(image):
     snr_text = None 
     snr_value = estimate_sigma(cv2.cvtColor(image,cv2.COLOR_RGB2GRAY), average_sigmas=False)
     logging.info(snr_value)
     if snr_value > 1 :
-        snr_text = "Signal to Noise is greater than 1 - More Signal in image "
+        snr_text = "SnR is greater than 1 meaning : image has less noise "
     else:
-        snr_text = "Signal to Noise is less than 1 - More Noise in image "
+        snr_text = "SnR is less than 1 Meaning : image has noise , it needs to be processed . "
     return snr_text