pro1.9.py

# -*- coding: utf-8 -*-
"""
Created on Thu Jun  7 16:30:19 2018

@author: Jui Shah
"""

#Road-detection from a traffic video-cam feed


import numpy as np
import cv2
import matplotlib.pyplot as plt

cap = cv2.VideoCapture('roadvideo2Trim.mp4')

while(cap.isOpened()):
    #frame by frame of video
    ret, image = cap.read()
       
    #creating empty image of same size
    height, width, no_use = image.shape
    empty_img = np.zeros((height, width), np.uint8)

    
    
    #APPLIED K-MEANS CLUSTERING
    Z = image.reshape((-1,3))
    # convert to np.float32
    Z = np.float32(Z)
    # define criteria, number of clusters(K) and apply kmeans()
    criteria = (cv2.TERM_CRITERIA_EPS + cv2.TERM_CRITERIA_MAX_ITER, 10, 1.0)
    K = 6
    ret,label,center=cv2.kmeans(Z,K,None,criteria,15,cv2.KMEANS_RANDOM_CENTERS)
    # Now convert back into uint8, and make original image
    center = np.uint8(center)
    res = center[label.flatten()]
    res2 = res.reshape((image.shape))
    
    
    
    #CONVERTED TO A LUV IMAGE AND MADE EMPTY IMAGE, A MASK
    blur = cv2.GaussianBlur(res2,(15,15),0)
    gray = cv2.cvtColor(blur,cv2.COLOR_RGB2GRAY)
    LUV = cv2.cvtColor(blur,cv2.COLOR_RGB2LUV)
    l = LUV[:,:,0]
    v1 = l>80
    v2 = l<150
    value_final = v1 & v2 
    empty_img[value_final] = 255
    empty_img[LUV[:,:100,:]] = 0
    
    
    #APPLIED BITWISE-AND ON GRAYSCALE IMAGE AND EMPTY IMAGE TO OBTAIN ROAD AND SOME-OTHER IMAGES TOO
    final = cv2.bitwise_and(gray,empty_img)
    final, contours, hierchary = cv2.findContours(final, cv2.RETR_TREE, cv2.CHAIN_APPROX_SIMPLE)
    final = cv2.drawContours(final, contours, -1, 0, 3)
    
    
    
    #FURTHER MASKED THE FINAL IMAGE TO OBTAIN ONLY THE ROAD PARTICLES
    final_masked = np.zeros((height, width), np.uint8)
    v1 = final >=91
    v2 = final <=130
    #v3 = final == 78    
    final_masked[v1 & v2] = 255
    
    
    #APPLIED EROSION,CONTOURS AND TOP-HAT TO REDUCE NOISE
    kernel = np.ones((3,3),np.uint8)
    final_eroded = cv2.erode(final_masked,kernel,iterations=1) 
    final_eroded, contours, hierchary = cv2.findContours(final_eroded, cv2.RETR_TREE, cv2.CHAIN_APPROX_SIMPLE)
    final_masked = cv2.drawContours(final_eroded, contours, -1, 0, 3)
    
    final_waste = cv2.morphologyEx(final_masked,cv2.MORPH_TOPHAT,kernel, iterations = 2) 
    final_waste = cv2.bitwise_not(final_waste)
    final_masked = cv2.bitwise_and(final_waste,final_masked)
    
    
    
    #MADE A LINE ON THE LEFT-BOTTOM OF THE PAGE
    final_masked = cv2.line(final_masked,(40,height),(400,height),255,100)
    #final_masked = cv2.line(final_masked,(width-300,height),(width,height),255,70)

    
    
    #USED FLOOD-FILL TO FILL IN THE SMALL BLACK LANES
    final_flood = final_masked.copy()
    h, w = final_masked.shape[:2]
    mask = np.zeros((h+2, w+2), np.uint8)
    cv2.floodFill(final_flood,mask,(0,0),255)
    final_flood = cv2.bitwise_not(final_flood)
    final_filled= cv2.bitwise_or(final_masked,final_flood)
    
    
    #final_blurred = cv2.GaussianBlur(final_filled,(5,5),0)
    
    cv2.namedWindow('original', cv2.WINDOW_NORMAL)
    cv2.imshow('original',image)
    cv2.namedWindow('tried_extraction', cv2.WINDOW_NORMAL)
    cv2.imshow('tried_extraction',final_filled)
    
    
    if cv2.waitKey(1) & 0xFF == ord('s'):
        break

cap.release()
cv2.destroyAllWindows()