基本阈值处理操作

本教程中你将学到如何：

• 使用 OpenCV 函数 cv::threshold 执行基本的阈值处理（thresholding）操作。

理论

注意： 以下说明出自 Bradski 与 Kaehler 所著的 Learning OpenCV 一书。

什么是阈值处理？

• 最简单的分割方法。
• 应用示例：将图像中与我们要分析的目标相对应的区域分离出来。这种分离基于目标像素与背景像素之间的强度变化。
• 为了把我们感兴趣的像素与其余像素（最终会被丢弃）区分开，我们将每个像素的强度值与一个阈值（threshold，根据要解决的问题来确定）进行比较。
• 一旦正确地分离出了重要的像素，我们就可以为它们设定一个确定的值来标识它们（即可以把它们赋值为 $0$（黑）、$255$（白）或任何符合你需求的值）。

阈值处理的类型

• OpenCV 提供了函数 cv::threshold 来执行阈值处理操作。
• 用该函数可以实现 $5$ 种类型的阈值处理操作，我们将在下面的小节中逐一说明。
• 为了说明这些阈值处理过程的工作原理，假设我们有一幅源图像，其像素强度值为 $src(x,y)$。在下方的示意图中，水平蓝线代表阈值 $thresh$（固定值）。

二值阈值处理（Threshold Binary）

• 该阈值处理操作可以表示为：

  $\texttt{dst}(x,y) = \begin{cases} \texttt{maxVal} & \text{if } \texttt{src}(x,y) > \texttt{thresh} \\ 0 & \text{otherwise} \end{cases}$

• 因此，如果像素 $src(x,y)$ 的强度高于 $thresh$，则新像素的强度被设为 $maxVal$；否则像素被设为 $0$。

反二值阈值处理（Threshold Binary, Inverted）

• 该阈值处理操作可以表示为：

  $\texttt{dst}(x,y) = \begin{cases} 0 & \text{if } \texttt{src}(x,y) > \texttt{thresh} \\ \texttt{maxVal} & \text{otherwise} \end{cases}$

• 如果像素 $src(x,y)$ 的强度高于 $thresh$，则新像素的强度被设为 $0$；否则被设为 $maxVal$。

截断（Truncate）

• 该阈值处理操作可以表示为：

  $\texttt{dst}(x,y) = \begin{cases} \texttt{thresh} & \text{if } \texttt{src}(x,y) > \texttt{thresh} \\ \texttt{src}(x,y) & \text{otherwise} \end{cases}$

• 像素的最大强度值为 $thresh$；如果 $src(x,y)$ 更大，则其值被截断。

阈值化为零（Threshold to Zero）

• 该操作可以表示为：

  $\texttt{dst}(x,y) = \begin{cases} \texttt{src}(x,y) & \text{if } \texttt{src}(x,y) > \texttt{thresh} \\ 0 & \text{otherwise} \end{cases}$

• 如果 $src(x,y)$ 低于 $thresh$，则新像素值将被设为 $0$。

反阈值化为零（Threshold to Zero, Inverted）

• 该操作可以表示为：

  $\texttt{dst}(x,y) = \begin{cases} 0 & \text{if } \texttt{src}(x,y) > \texttt{thresh} \\ \texttt{src}(x,y) & \text{otherwise} \end{cases}$

• 如果 $src(x,y)$ 大于 $thresh$，则新像素值将被设为 $0$。

代码

本教程的代码如下所示。你也可以从 这里 下载 C++ 版本。

/**
 * @file Threshold.cpp
 * @brief Sample code that shows how to use the diverse threshold options offered by OpenCV
 * @author OpenCV team
 */

#include "opencv2/imgproc.hpp"
#include "opencv2/imgcodecs.hpp"
#include "opencv2/highgui.hpp"
#include <iostream>

using namespace cv;
using std::cout;

/// Global variables

int threshold_value = 0;
int threshold_type = 3;
int const max_value = 255;
int const max_type = 4;
int const max_binary_value = 255;

Mat src, src_gray, dst;
const char* window_name = "Threshold Demo";

const char* trackbar_type = "Type: \n 0: Binary \n 1: Binary Inverted \n 2: Truncate \n 3: To Zero \n 4: To Zero Inverted";
const char* trackbar_value = "Value";

/**
 * @function Threshold_Demo
 */
static void Threshold_Demo( int, void* )
{
    /* 0: Binary
     1: Binary Inverted
     2: Threshold Truncated
     3: Threshold to Zero
     4: Threshold to Zero Inverted
    */
    threshold( src_gray, dst, threshold_value, max_binary_value, threshold_type );
    imshow( window_name, dst );
}

/**
 * @function main
 */
int main( int argc, char** argv )
{
    String imageName("stuff.jpg"); // by default
    if (argc > 1)
    {
        imageName = argv[1];
    }
    src = imread( samples::findFile( imageName ), IMREAD_COLOR ); // Load an image

    if (src.empty())
    {
        cout << "Cannot read the image: " << imageName << std::endl;
        return -1;
    }

    cvtColor( src, src_gray, COLOR_BGR2GRAY ); // Convert the image to Gray

    namedWindow( window_name, WINDOW_AUTOSIZE ); // Create a window to display results

    createTrackbar( trackbar_type,
                    window_name, &threshold_type,
                    max_type, Threshold_Demo ); // Create a Trackbar to choose type of Threshold

    createTrackbar( trackbar_value,
                    window_name, &threshold_value,
                    max_value, Threshold_Demo ); // Create a Trackbar to choose Threshold value

    Threshold_Demo( 0, 0 ); // Call the function to initialize

    /// Wait until the user finishes the program
    waitKey();
    return 0;
}

Java 版本可从 这里 下载。

import java.awt.BorderLayout;
import java.awt.Container;
import java.awt.Image;

import javax.swing.BoxLayout;
import javax.swing.ImageIcon;
import javax.swing.JFrame;
import javax.swing.JLabel;
import javax.swing.JPanel;
import javax.swing.JSlider;
import javax.swing.event.ChangeEvent;
import javax.swing.event.ChangeListener;

import org.opencv.core.Core;
import org.opencv.core.Mat;
import org.opencv.highgui.HighGui;
import org.opencv.imgcodecs.Imgcodecs;
import org.opencv.imgproc.Imgproc;

public class Threshold {
    private static int MAX_VALUE = 255;
    private static int MAX_TYPE = 4;
    private static int MAX_BINARY_VALUE = 255;
    private static final String WINDOW_NAME = "Threshold Demo";
    private static final String TRACKBAR_TYPE = "<html><body>Type: <br> 0: Binary <br> "
            + "1: Binary Inverted <br> 2: Truncate <br> "
            + "3: To Zero <br> 4: To Zero Inverted</body></html>";
    private static final String TRACKBAR_VALUE = "Value";
    private int thresholdValue = 0;
    private int thresholdType = 3;
    private Mat src;
    private Mat srcGray = new Mat();
    private Mat dst = new Mat();
    private JFrame frame;
    private JLabel imgLabel;

    public Threshold(String[] args) {
        String imagePath = "../data/stuff.jpg";
        if (args.length > 0) {
            imagePath = args[0];
        }
        // Load an image
        src = Imgcodecs.imread(imagePath);
        if (src.empty()) {
            System.out.println("Empty image: " + imagePath);
            System.exit(0);
        }
        // Convert the image to Gray
        Imgproc.cvtColor(src, srcGray, Imgproc.COLOR_BGR2GRAY);

        // Create and set up the window.
        frame = new JFrame(WINDOW_NAME);
        frame.setDefaultCloseOperation(JFrame.EXIT_ON_CLOSE);
        // Set up the content pane.
        Image img = HighGui.toBufferedImage(srcGray);
        addComponentsToPane(frame.getContentPane(), img);
        // Use the content pane's default BorderLayout. No need for
        // setLayout(new BorderLayout());
        // Display the window.
        frame.pack();
        frame.setVisible(true);
    }

    private void addComponentsToPane(Container pane, Image img) {
        if (!(pane.getLayout() instanceof BorderLayout)) {
            pane.add(new JLabel("Container doesn't use BorderLayout!"));
            return;
        }

        JPanel sliderPanel = new JPanel();
        sliderPanel.setLayout(new BoxLayout(sliderPanel, BoxLayout.PAGE_AXIS));

        sliderPanel.add(new JLabel(TRACKBAR_TYPE));
        // Create Trackbar to choose type of Threshold
        JSlider sliderThreshType = new JSlider(0, MAX_TYPE, thresholdType);
        sliderThreshType.setMajorTickSpacing(1);
        sliderThreshType.setMinorTickSpacing(1);
        sliderThreshType.setPaintTicks(true);
        sliderThreshType.setPaintLabels(true);
        sliderPanel.add(sliderThreshType);

        sliderPanel.add(new JLabel(TRACKBAR_VALUE));
        // Create Trackbar to choose Threshold value
        JSlider sliderThreshValue = new JSlider(0, MAX_VALUE, 0);
        sliderThreshValue.setMajorTickSpacing(50);
        sliderThreshValue.setMinorTickSpacing(10);
        sliderThreshValue.setPaintTicks(true);
        sliderThreshValue.setPaintLabels(true);
        sliderPanel.add(sliderThreshValue);

        sliderThreshType.addChangeListener(new ChangeListener() {
            @Override
            public void stateChanged(ChangeEvent e) {
                JSlider source = (JSlider) e.getSource();
                thresholdType = source.getValue();
                update();
            }
        });

        sliderThreshValue.addChangeListener(new ChangeListener() {
            @Override
            public void stateChanged(ChangeEvent e) {
                JSlider source = (JSlider) e.getSource();
                thresholdValue = source.getValue();
                update();
            }
        });

        pane.add(sliderPanel, BorderLayout.PAGE_START);
        imgLabel = new JLabel(new ImageIcon(img));
        pane.add(imgLabel, BorderLayout.CENTER);
    }

    private void update() {
        Imgproc.threshold(srcGray, dst, thresholdValue, MAX_BINARY_VALUE, thresholdType);
        Image img = HighGui.toBufferedImage(dst);
        imgLabel.setIcon(new ImageIcon(img));
        frame.repaint();
    }

    public static void main(String[] args) {
        // Load the native OpenCV library
        System.loadLibrary(Core.NATIVE_LIBRARY_NAME);

        // Schedule a job for the event dispatch thread:
        // creating and showing this application's GUI.
        javax.swing.SwingUtilities.invokeLater(new Runnable() {
            @Override
            public void run() {
                new Threshold(args);
            }
        });
    }
}

Python 版本可从 这里 下载。

from __future__ import print_function
import cv2 as cv
import argparse

max_value = 255
max_type = 4
max_binary_value = 255
trackbar_type = 'Type: \n 0: Binary \n 1: Binary Inverted \n 2: Truncate \n 3: To Zero \n 4: To Zero Inverted'
trackbar_value = 'Value'
window_name = 'Threshold Demo'

def Threshold_Demo(val):
    #0: Binary
    #1: Binary Inverted
    #2: Threshold Truncated
    #3: Threshold to Zero
    #4: Threshold to Zero Inverted
    threshold_type = cv.getTrackbarPos(trackbar_type, window_name)
    threshold_value = cv.getTrackbarPos(trackbar_value, window_name)
    _, dst = cv.threshold(src_gray, threshold_value, max_binary_value, threshold_type )
    cv.imshow(window_name, dst)

parser = argparse.ArgumentParser(description='Code for Basic Thresholding Operations tutorial.')
parser.add_argument('--input', help='Path to input image.', default='stuff.jpg')
args = parser.parse_args()

# Load an image
src = cv.imread(cv.samples.findFile(args.input))
if src is None:
    print('Could not open or find the image: ', args.input)
    exit(0)
# Convert the image to Gray
src_gray = cv.cvtColor(src, cv.COLOR_BGR2GRAY)

# Create a window to display results
cv.namedWindow(window_name)

# Create Trackbar to choose type of Threshold
cv.createTrackbar(trackbar_type, window_name , 3, max_type, Threshold_Demo)
# Create Trackbar to choose Threshold value
cv.createTrackbar(trackbar_value, window_name , 0, max_value, Threshold_Demo)

# Call the function to initialize
Threshold_Demo(0)
# Wait until user finishes program
cv.waitKey()

代码解析

我们来看看程序的总体结构：

1. 载入一幅图像。如果它是 BGR 格式，则将其转换为灰度图。为此，记住我们可以使用函数 cv::cvtColor：

   String imageName("stuff.jpg"); // by default
   if (argc > 1)
   {
       imageName = argv[1];
   }
   src = imread( samples::findFile( imageName ), IMREAD_COLOR ); // Load an image
   
   if (src.empty())
   {
       cout << "Cannot read the image: " << imageName << std::endl;
       return -1;
   }
   
   cvtColor( src, src_gray, COLOR_BGR2GRAY ); // Convert the image to Gray

```java
String imagePath = "../data/stuff.jpg";
if (args.length > 0) {
    imagePath = args[0];
}
// Load an image
src = Imgcodecs.imread(imagePath);
if (src.empty()) {
    System.out.println("Empty image: " + imagePath);
    System.exit(0);
}
// Convert the image to Gray
Imgproc.cvtColor(src, srcGray, Imgproc.COLOR_BGR2GRAY);
```

```python
# Load an image
src = cv.imread(cv.samples.findFile(args.input))
if src is None:
    print('Could not open or find the image: ', args.input)
    exit(0)
# Convert the image to Gray
src_gray = cv.cvtColor(src, cv.COLOR_BGR2GRAY)
```

1. 创建一个窗口以显示结果：

   namedWindow( window_name, WINDOW_AUTOSIZE ); // Create a window to display results

   // Create and set up the window.
   frame = new JFrame(WINDOW_NAME);
   frame.setDefaultCloseOperation(JFrame.EXIT_ON_CLOSE);
   // Set up the content pane.
   Image img = HighGui.toBufferedImage(srcGray);
   addComponentsToPane(frame.getContentPane(), img);
   // Use the content pane's default BorderLayout. No need for
   // setLayout(new BorderLayout());
   // Display the window.
   frame.pack();
   frame.setVisible(true);

   # Create a window to display results
   cv.namedWindow(window_name)

2. 创建 $2$ 个滑动条（Trackbar）供用户输入：

   • 阈值处理类型：二值、化为零等等……
   • 阈值

   createTrackbar( trackbar_type,
                   window_name, &threshold_type,
                   max_type, Threshold_Demo ); // Create a Trackbar to choose type of Threshold
   
   createTrackbar( trackbar_value,
                   window_name, &threshold_value,
                   max_value, Threshold_Demo ); // Create a Trackbar to choose Threshold value

   sliderPanel.add(new JLabel(TRACKBAR_TYPE));
   // Create Trackbar to choose type of Threshold
   JSlider sliderThreshType = new JSlider(0, MAX_TYPE, thresholdType);
   sliderThreshType.setMajorTickSpacing(1);
   sliderThreshType.setMinorTickSpacing(1);
   sliderThreshType.setPaintTicks(true);
   sliderThreshType.setPaintLabels(true);
   sliderPanel.add(sliderThreshType);
   
   sliderPanel.add(new JLabel(TRACKBAR_VALUE));
   // Create Trackbar to choose Threshold value
   JSlider sliderThreshValue = new JSlider(0, MAX_VALUE, 0);
   sliderThreshValue.setMajorTickSpacing(50);
   sliderThreshValue.setMinorTickSpacing(10);
   sliderThreshValue.setPaintTicks(true);
   sliderThreshValue.setPaintLabels(true);
   sliderPanel.add(sliderThreshValue);

   # Create Trackbar to choose type of Threshold
   cv.createTrackbar(trackbar_type, window_name , 3, max_type, Threshold_Demo)
   # Create Trackbar to choose Threshold value
   cv.createTrackbar(trackbar_value, window_name , 0, max_value, Threshold_Demo)

3. 等待用户输入阈值与阈值处理类型（或直到程序退出）。

4. 每当用户改变任一滑动条的值时，函数 Threshold_Demo（Java 中为 update）就会被调用：

   static void Threshold_Demo( int, void* )
   {
       /* 0: Binary
        1: Binary Inverted
        2: Threshold Truncated
        3: Threshold to Zero
        4: Threshold to Zero Inverted
       */
       threshold( src_gray, dst, threshold_value, max_binary_value, threshold_type );
       imshow( window_name, dst );
   }

   private void update() {
       Imgproc.threshold(srcGray, dst, thresholdValue, MAX_BINARY_VALUE, thresholdType);
       Image img = HighGui.toBufferedImage(dst);
       imgLabel.setIcon(new ImageIcon(img));
       frame.repaint();
   }

   def Threshold_Demo(val):
       #0: Binary
       #1: Binary Inverted
       #2: Threshold Truncated
       #3: Threshold to Zero
       #4: Threshold to Zero Inverted
       threshold_type = cv.getTrackbarPos(trackbar_type, window_name)
       threshold_value = cv.getTrackbarPos(trackbar_value, window_name)
       _, dst = cv.threshold(src_gray, threshold_value, max_binary_value, threshold_type )
       cv.imshow(window_name, dst)

可以看到，这里调用了函数 cv::threshold。我们在 C++ 代码中给它传了 $5$ 个参数：

• src_gray：我们的输入图像
• dst：目标（输出）图像
• threshold_value：阈值处理操作所参照的 $thresh$ 值
• max_BINARY_value：二值阈值处理操作中使用的值（用于设定被选中的像素）
• threshold_type：$5$ 种阈值处理操作之一。它们列在上面函数的注释中。

结果

1. 编译该程序后，运行时把一幅图像的路径作为参数传入。

2. 首先，我们尝试用反二值阈值处理图像。我们期望比 $thresh$ 亮的像素会变暗，实际情况正是如此（从原图可以注意到，小狗的舌头和眼睛相比图像其他部分特别亮，这一点反映在输出图像中）。

3. 现在我们尝试阈值化为零。我们期望最暗的像素（低于阈值）将变为全黑，而值大于阈值的像素将保持其原始值，输出图像的快照证实了这一点。