7

I am using following code to detect edges from given document.

private Mat edgeDetection(Mat src) {
    Mat edges = new Mat();
    Imgproc.cvtColor(src, edges, Imgproc.COLOR_BGR2GRAY);
    Imgproc.GaussianBlur(edges, edges, new Size(5, 5), 0);
    Imgproc.Canny(edges, edges, 10, 30);
    return edges;
}

And then I can find the document from this edges by finding largest contour from this.

My problem is I can find the document from following pic:

enter image description here

but not from following pic:

enter image description here

How can I improve this edge detection?

halfer
  • 18,701
  • 13
  • 79
  • 158
Gunaseelan
  • 11,546
  • 9
  • 64
  • 109

2 Answers2

10

I use Python, but the main idea is the same.

If you directly do cvtColor: bgr -> gray for img2, then you must fail. Because the gray becames difficulty to distinguish the regions:

enter image description here

Related answers:

  1. How to detect colored patches in an image using OpenCV?
  2. Edge detection on colored background using OpenCV
  3. OpenCV C++/Obj-C: Detecting a sheet of paper / Square Detection

In your image, the paper is white, while the background is colored. So, it's better to detect the paper is Saturation(饱和度) channel in HSV color space. For HSV, refer to https://en.wikipedia.org/wiki/HSL_and_HSV#Saturation.

Main steps:

  1. Read into BGR
  2. Convert the image from bgr to hsv space
  3. Threshold the S channel
  4. Then find the max external contour(or do Canny, or HoughLines as you like, I choose findContours), approx to get the corners.

This is the first result:

enter image description here

This is the second result:

enter image description here

The Python code(Python 3.5 + OpenCV 3.3):

#!/usr/bin/python3
# 2017.12.20 10:47:28 CST
# 2017.12.20 11:29:30 CST

import cv2
import numpy as np

##(1) read into  bgr-space
img = cv2.imread("test2.jpg")

##(2) convert to hsv-space, then split the channels
hsv = cv2.cvtColor(img, cv2.COLOR_BGR2HSV)
h,s,v = cv2.split(hsv)

##(3) threshold the S channel using adaptive method(`THRESH_OTSU`) or fixed thresh
th, threshed = cv2.threshold(s, 50, 255, cv2.THRESH_BINARY_INV)

##(4) find all the external contours on the threshed S
cnts = cv2.findContours(threshed, cv2.RETR_EXTERNAL, cv2.CHAIN_APPROX_SIMPLE)[-2]
canvas  = img.copy()
#cv2.drawContours(canvas, cnts, -1, (0,255,0), 1)

## sort and choose the largest contour
cnts = sorted(cnts, key = cv2.contourArea)
cnt = cnts[-1]

## approx the contour, so the get the corner points
arclen = cv2.arcLength(cnt, True)
approx = cv2.approxPolyDP(cnt, 0.02* arclen, True)
cv2.drawContours(canvas, [cnt], -1, (255,0,0), 1, cv2.LINE_AA)
cv2.drawContours(canvas, [approx], -1, (0, 0, 255), 1, cv2.LINE_AA)

## Ok, you can see the result as tag(6)
cv2.imwrite("detected.png", canvas)
Kinght 金
  • 14,440
  • 4
  • 49
  • 62
  • May you explain what are the `th` and `threshold` in this line of your code: `th, threshed = cv2.threshold(s, 50, 255, cv2.THRESH_BINARY_INV)`. I am trying to convert and test your code in c++ but don't know how to do it in this line? As I searched, the threshold func has only one double output but it seems you get two of them(although I don't know python and only guessed) – Hasani Dec 31 '17 at 08:58
  • 2
    @Hasani `th` is the returned thresh value; `threshed` is the threshold binary image. The function prototype: cv2.threshold(src, thresh, maxval, type[, dst]) -> retval, dst – Kinght 金 Dec 31 '17 at 09:41
1

In OpenCV there is function called dilate this will darker the lines. so try the code like below.

private Mat edgeDetection(Mat src) {
    Mat edges = new Mat();
    Imgproc.cvtColor(src, edges, Imgproc.COLOR_BGR2GRAY);
    Imgproc.dilate(edges, edges, Imgproc.getStructuringElement(Imgproc.MORPH_RECT, new Size(10, 10)));
    Imgproc.GaussianBlur(edges, edges, new Size(5, 5), 0);
    Imgproc.Canny(edges, edges, 15, 15 * 3);
    return edges;
}
Prakash
  • 730
  • 5
  • 15