问题描述
我认为这个问题是之前提出的，但是我没有找到解决问题的样本或解决方案.我是opencv的新手，我想使用OpenCV CameraPreview进行纸张检测.在我的示例应用程序中，我将opencv 3.0.0与静态初始化一起使用.我了解对象识别可以通过以下步骤完成:
I think this question is asked before but i didnt find a sample or solution for my problem. I am new to opencv and i want to use the OpenCV CameraPreview for paper sheet detection. In my sample app i use opencv 3.0.0 with static initialization.I understand that object recognition can done with these steps:
- 将输入图像设为Canny
- 模糊Canny图片
- 在模糊的Canny图像上找到轮廓
- 搜索矩形等
- 绘制线条或用半透明颜色填充矩形
我的问题是我现在无法对图像进行模糊处理，但是我不知道如何找到轮廓和矩形并用半透明颜色填充它们.
My problem is now that i can canny and blur the image but i dont know how to find contours and rectangles and fill them with a half transparent color.
这是我当前的onCameraFrame函数:
Here is my current onCameraFrame Function:
@Override
public Mat onCameraFrame(CameraBridgeViewBase.CvCameraViewFrame inputFrame) {
Mat input = inputFrame.rgba();
Mat output = input.clone();
Imgproc.Canny(input, output, 50, 50);
Imgproc.blur(output, output,new Size(5,5));
//Find Contours
//Search for biggest Contour/Rectangle
//Fill Rectangle with half transparent Color
return output;
}
有人可以帮助我解决纸张检测问题，并且有适用于android/java的代码示例吗?谢谢
Can anybody help me to solve the problem of paper sheet detection and has a code sample for android/java?Thank you
推荐答案
以下代码来自打开便笺扫描器我正在开发的应用程序，您可以使用它来查找更多信息.
The following code is from the Open Note Scanner app I am developing, you can use it to look for more information.
函数findDocument将返回一个四边形对象，该对象封装了带有轮廓的MatOfPoint和带有单个点的Point [].您可以调用它，并使用返回的对象调用Imgproc.drawContours()完成图像.
the function findDocument will return a Quadrilateral object that encapsulates a MatOfPoint with the contour and a Point[] with individual points. You can call it and with the returned object call the Imgproc.drawContours() to finish your image.
所有代码均以来自pyimagesearch的出色教程
注意:这是从我的代码中快速移植方法的过程，它没有语法错误，但是我没有对其进行测试.
NOTICE: this was a quick transplant of the methods from my code, it doesn't have syntax errors but I didn't tested it.
package com.todobom.opennotescanner.views;
import org.opencv.core.CvType;
import org.opencv.core.Mat;
import org.opencv.core.MatOfPoint;
import org.opencv.core.MatOfPoint2f;
import org.opencv.core.Point;
import org.opencv.core.Size;
import org.opencv.imgproc.Imgproc;
import java.util.ArrayList;
import java.util.Arrays;
import java.util.Collections;
import java.util.Comparator;
public class detectDocument {
/**
* Object that encapsulates the contour and 4 points that makes the larger
* rectangle on the image
*/
public static class Quadrilateral {
public MatOfPoint contour;
public Point[] points;
public Quadrilateral(MatOfPoint contour, Point[] points) {
this.contour = contour;
this.points = points;
}
}
public static Quadrilateral findDocument( Mat inputRgba ) {
ArrayList<MatOfPoint> contours = findContours(inputRgba);
Quadrilateral quad = getQuadrilateral(contours);
return quad;
}
private static ArrayList<MatOfPoint> findContours(Mat src) {
double ratio = src.size().height / 500;
int height = Double.valueOf(src.size().height / ratio).intValue();
int width = Double.valueOf(src.size().width / ratio).intValue();
Size size = new Size(width,height);
Mat resizedImage = new Mat(size, CvType.CV_8UC4);
Mat grayImage = new Mat(size, CvType.CV_8UC4);
Mat cannedImage = new Mat(size, CvType.CV_8UC1);
Imgproc.resize(src,resizedImage,size);
Imgproc.cvtColor(resizedImage, grayImage, Imgproc.COLOR_RGBA2GRAY, 4);
Imgproc.GaussianBlur(grayImage, grayImage, new Size(5, 5), 0);
Imgproc.Canny(grayImage, cannedImage, 75, 200);
ArrayList<MatOfPoint> contours = new ArrayList<MatOfPoint>();
Mat hierarchy = new Mat();
Imgproc.findContours(cannedImage, contours, hierarchy, Imgproc.RETR_LIST, Imgproc.CHAIN_APPROX_SIMPLE);
hierarchy.release();
Collections.sort(contours, new Comparator<MatOfPoint>() {
@Override
public int compare(MatOfPoint lhs, MatOfPoint rhs) {
return Double.valueOf(Imgproc.contourArea(rhs)).compareTo(Imgproc.contourArea(lhs));
}
});
resizedImage.release();
grayImage.release();
cannedImage.release();
return contours;
}
private static Quadrilateral getQuadrilateral(ArrayList<MatOfPoint> contours) {
for ( MatOfPoint c: contours ) {
MatOfPoint2f c2f = new MatOfPoint2f(c.toArray());
double peri = Imgproc.arcLength(c2f, true);
MatOfPoint2f approx = new MatOfPoint2f();
Imgproc.approxPolyDP(c2f, approx, 0.02 * peri, true);
Point[] points = approx.toArray();
// select biggest 4 angles polygon
if (points.length == 4) {
Point[] foundPoints = sortPoints(points);
return new Quadrilateral(c, foundPoints);
}
}
return null;
}
private static Point[] sortPoints(Point[] src) {
ArrayList<Point> srcPoints = new ArrayList<>(Arrays.asList(src));
Point[] result = { null , null , null , null };
Comparator<Point> sumComparator = new Comparator<Point>() {
@Override
public int compare(Point lhs, Point rhs) {
return Double.valueOf(lhs.y + lhs.x).compareTo(rhs.y + rhs.x);
}
};
Comparator<Point> diffComparator = new Comparator<Point>() {
@Override
public int compare(Point lhs, Point rhs) {
return Double.valueOf(lhs.y - lhs.x).compareTo(rhs.y - rhs.x);
}
};
// top-left corner = minimal sum
result[0] = Collections.min(srcPoints, sumComparator);
// bottom-right corner = maximal sum
result[2] = Collections.max(srcPoints, sumComparator);
// top-right corner = minimal diference
result[1] = Collections.min(srcPoints, diffComparator);
// bottom-left corner = maximal diference
result[3] = Collections.max(srcPoints, diffComparator);
return result;
}
}
这篇关于Android OpenCV纸张检测的文章就介绍到这了，希望我们推荐的答案对大家有所帮助，也希望大家多多支持！