#Import Required Packages
import numpy as np
import gradio as gr
#from google.colab.patches import cv2_imshow
import cv2
import matplotlib.pyplot as plt
import numpy as np
import skimage 
import imutils
from imutils import contours 
def figplota(xvalues):
  fig = plt.figure()
  plt.plot(xvalues, figure=fig)
  return fig
def quant(imageinput):
  #@title Please Input the Lateral Flow Assay Image
  # read image using openCV
  #path = "/content/l1.jpg"
  image = cv2.imread(imageinput)#imageinput
  target = "PKU" 
  #print(image)
  #cv2_imshow(image)
  # Convert the image to grayscale 
  BGR2RGB = cv2.cvtColor(image, cv2.COLOR_BGR2RGB)
  gray = cv2.cvtColor(BGR2RGB, cv2.COLOR_RGB2GRAY)
  #print(gray)
  #cv2_imshow(gray)
  # Invert the image to negative scale
  negative = cv2.bitwise_not(gray)
  negativeimage = negative.copy() #save a copy to avoid disrupting the image contour
  #print(negativeimage)
  #cv2_imshow(negativeimage)
  # Minimize the noisy effects of artificats using Gaussian blur (helps with minimizing the effect of noisy artifactual bright-spots)
  blur = cv2.GaussianBlur(negativeimage, (11, 11), 0)
  #print(blur)
  #cv2_imshow(blur)
  # Binarize Image
  threshold = float(cv2.meanStdDev(blur)[0]) + 0.6*float(cv2.meanStdDev(blur)[1])
  imgthreshold = cv2.threshold(blur, threshold, 255, cv2.THRESH_BINARY)[1]
  #print(imgthreshold)
  #cv2_imshow(image_thresh)
  # Reducing noise noise through eroding & eroding 
  imgeroding = cv2.erode(imgthreshold, None, iterations=1)
  zeronoise = cv2.dilate(imgeroding, None, iterations=1)
  #print(zeronoise)
  #cv2_imshow(zeronoise)
  # CCA the threshold Image
  import skimage.measure
  labels = skimage.measure.label(zeronoise, background=0)
  masking = np.zeros(zeronoise.shape, dtype="uint8")
  for label in np.unique(labels):
      if label == 0:
          continue
      MaskL = np.zeros(zeronoise.shape, dtype="uint8")
      MaskL[labels == label] = 255
      numPixels = cv2.countNonZero(MaskL)
      if numPixels > masking.shape[1]*3:
          masking = cv2.add(masking, MaskL)    
  #cv2_imshow(mask)
  # Find the contours and sort, please change from bottom-to-top to top-to-bottom accordingly
  contourss = cv2.findContours(masking.copy(), cv2.RETR_EXTERNAL,
      cv2.CHAIN_APPROX_SIMPLE)
  contourss = imutils.grab_contours(contourss)
  contourss = contours.sort_contours(contourss, method="bottom-to-top")[0] #change here accordingly
  final= []
  if len(contourss) > 1:
      for (i, c) in enumerate(contourss):
          # draw the bright spot on the image for the control and sample band 
          x, y, width, height = cv2.boundingRect(c)
          final.append(negativeimage[y:y+height, x:x+width])
          rect = cv2.minAreaRect(c)
          box = cv2.boxPoints(rect)
          # convert all coordinates floating point values to int
          box = np.int0(box)
          # draw a rectangle
          cv2.drawContours(image, [box], 0, (0, 0, 255), thickness=2)

  elif len(contourss) == 1:
          # draw the bright spot on the image for the control band
          for (i, c) in enumerate(contourss):
              x, y, width, height = cv2.boundingRect(c)
              final.append(negativeimage[y:y+height, x:x+width])
              rect = cv2.minAreaRect(c)
              box = cv2.boxPoints(rect)
              # convert all coordinates floating point values to int
              box = np.int0(box)
              # draw a rectangle
              cv2.drawContours(image, [box], 0, (0, 0, 255), thickness=2)

          box_ctl_final = np.array([[0,height], [0,0], [width,0], [width,height]])
          final_test_1 = negativeimage[height:2*height, 0:width]
          final_test_2 = negativeimage[2*height:3*height, 0:width]
          
          if cv2.meanStdDev(final_test_1)[0] > cv2.meanStdDev(final_test_2)[0]:
              box_ctl_final[:,1] = box_ctl_final[:,1]+height
              final.append(final_test_1)
          else:
              box_ctl_final[:,1] = box_ctl_final[:,1]+2*height
              final.append(final_test_2)
          
          cv2.drawContours(image, [box_ctl_final], 0, (0, 0, 255), thickness=2)
          
  # Return error message for unclear tests
  else :
      print("No Bands Detected")
  #print(image)
  #cv2_imshow(image)
  # generate signal ratio of sample to control band, you can change according to sorting of bands

  ratio = float(cv2.meanStdDev(final[1])[0] / cv2.meanStdDev(final[0])[0])
  thresho = 0.50 
  sig=(final[1][0]/final[0][0])
  #signal=plt.plot(sig,figure=plt.figure())
  

  if ratio >= thresho:
    xx=str("The test band signal [" + str(ratio) + "] shows a " + target +"-POSITIVE test.")
  else:
    xx=str("The test band signal[" + str(ratio) + "] shows a " + target +"-NEGATIVE test.")
  
  return xx,figplota(sig),cv2.resize(image, (20,60), interpolation = cv2.INTER_AREA) #cv2.resize(signal, (20,40), interpolation = cv2.INTER_AREA)#,cv2.resize(signal, (20,40), interpolation = cv2.INTER_AREA)
iface  = gr.Interface(quant, gr.Image(type="filepath"),  outputs=["text","plot","image"])
iface.launch()