moneydetector.py

import numpy as np
import cv2

def angle(a, b, c):
    return ((a - c).dot(np.transpose(a - c)) + (b - c).dot(np.transpose(b - c)) - (a - b).dot(np.transpose(a - b))) / (2 * (a - c).dot(np.transpose(b - c)))

def detect(dollar_image):

    squares = []

    #dollar_image = cv2.imread('7dollar.png', cv2.IMREAD_UNCHANGED)
    #dollar_image = cv2.medianBlur(dollar_image, 3)

    # Convert to grayscale
    dollar_image_gray = cv2.cvtColor(dollar_image, cv2.COLOR_BGR2GRAY)

    thresh = cv2.adaptiveThreshold(dollar_image_gray, 255, cv2.ADAPTIVE_THRESH_GAUSSIAN_C, cv2.THRESH_BINARY, 971, 2)

    im2, contours, hierarchy = cv2.findContours(thresh, cv2.RETR_LIST, cv2.CHAIN_APPROX_SIMPLE)


    if not contours:
        print "No contours!"
    else:
        # test each contour
        print len(contours)
        for contour in contours:
            # approximate contour with accuracy proportional
            # to the contour perimeter
            result = cv2.approxPolyDP(contour, cv2.arcLength(contour, True)*0.02, True)
            # square contours should have 4 vertices after approximation
            # relatively large area (to filter out noisy contours)
            # and be convex.
            # Note: absolute value of an area is used because
            # area may be positive or negative - in accordance with the
            # contour orientation
            if(len(result) >= 4 or (len(result) == 4 and 
                abs(cv2.contourArea(result)) > 1000 and
                cv2.isContourConvex(result) )):
            
                # print "four sided"
                s = 0;
                """for i in range(4):
                    # find minimum angle between joint
                    # edges (maximum of cosine)
                    t = abs(angle( result[i], result[i-2], result[i-1]))
                    if s<t:
                        s=t"""
                # if cosines of all angles are small
                # (all angles are ~90 degree) then write quandrange
                # vertices to resultant sequence
                if( s < 0.3 ):
                    for i in range(4):
                        squares.append( result )

    #print squares
    
    for square in squares:
        # print square
        cv2.rectangle(dollar_image, tuple(square[0][0]), tuple(square[2][0]), (255, 0, 0))

    screen_res = 1900, 1080
    scale_width = screen_res[0] / thresh.shape[1]
    scale_height = screen_res[1] / thresh.shape[0]
    scale = min(scale_width, scale_height)
    window_width = int(thresh.shape[1] * scale)
    window_height = int(thresh.shape[0] * scale)

    cv2.namedWindow("thresh.jpg", cv2.WINDOW_NORMAL)
    cv2.resizeWindow("thresh.jpg", window_width, window_height)

    cv2.imshow("thresh.jpg", thresh)
    cv2.waitKey()

    cv2.namedWindow("Hi.jpg", cv2.WINDOW_NORMAL)
    cv2.resizeWindow("Hi.jpg", window_width, window_height)

    cv2.imshow("Hi.jpg", dollar_image)
    cv2.waitKey()
#cv2.imwrite("thresh.jpg", thresh)
#cv2.imwrite("Hi.jpg", dollar_image)