最近有个需求，希望识别图片上的虫子，对于java来说，图像识别不是很好做。在网上也搜索了很多，很多的代码都是不完整，或者下载下载报错，有的写的很长看不懂。所以自己试着用java的opencv写了一段代码。发现识别的效果还不错，下面把代码贴出来。有需要的可以参考。但是这里面有一些缺陷，就是没有加入transformer和org.deeplearning4j，对于复杂的场景识别的不是很准确。后期再更新把神经网络加上去。
一、POM.xml文件导入jar包
<dependency><groupId>org.bytedeco</groupId><artifactId>javacv-platform</artifactId><version>1.5.10</version> <!-- 根据最新版本更新 --></dependency><dependency><groupId>org.opencv</groupId><artifactId>opencv</artifactId><version>4.9.0</version></dependency><dependency><groupId>org.deeplearning4j</groupId><artifactId>deeplearning4j-core</artifactId><version>1.0.0-M1.1</version></dependency><dependency><groupId>org.nd4j</groupId><artifactId>nd4j-native</artifactId><version>1.0.0-M2</version></dependency><dependency><groupId>org.nd4j</groupId><artifactId>nd4j-api</artifactId><version>1.0.0-M2</version></dependency>二、主要的处理步骤和逻辑代码
- 读取图片
- 截取区域
- 转换灰度
- 二值化
- 高斯
- 中指滤波
- 双边滤波
- 去除线框干扰
- 形态学操作
- 轮廓检测
package org.example;import org.bytedeco.opencv.global.opencv_imgcodecs;
import org.bytedeco.opencv.global.opencv_imgproc;
import org.bytedeco.opencv.opencv_core.*;public class BugCounterTest {public static void main(String[] args) {// 读取图片文件Mat src = opencv_imgcodecs.imread("C:\\Users\\HP\\Desktop\\aaaa.png");if (src.empty()) {System.out.println("Error: Cannot read image!");return;}// 截取感兴趣区域Rect roi = new Rect(0, 180, 1300, 600);Mat croppedImage = new Mat(src, roi);// 转换为灰度图像Mat gray = new Mat();opencv_imgproc.cvtColor(croppedImage, gray, opencv_imgproc.COLOR_BGR2GRAY);// 二值化图像Mat binary = new Mat();opencv_imgproc.threshold(gray, binary, 100, 255, opencv_imgproc.THRESH_BINARY_INV);//高斯模糊处理Mat blurredImage = new Mat();opencv_imgproc.GaussianBlur(binary,blurredImage,new Size(5, 5),0);//中值滤波Mat medianFilteredImage = new Mat();opencv_imgproc.medianBlur(blurredImage, medianFilteredImage, 5);// 双边滤波Mat bilateralFilteredImage = new Mat();opencv_imgproc.bilateralFilter(medianFilteredImage, bilateralFilteredImage, 9, 75, 75);// 去除线框干扰Mat edgeImage = new Mat();opencv_imgproc.Canny(bilateralFilteredImage, edgeImage, 50, 150); // 可调整参数// 形态学操作Mat kernel = opencv_imgproc.getStructuringElement(opencv_imgproc.MORPH_RECT, new Size(3, 3));opencv_imgproc.dilate(edgeImage, edgeImage, kernel);opencv_imgproc.erode(edgeImage, edgeImage, kernel);// 轮廓检测MatVector contours = new MatVector();Mat hierarchy = new Mat();opencv_imgproc.findContours(edgeImage, contours, hierarchy, opencv_imgproc.RETR_EXTERNAL, opencv_imgproc.CHAIN_APPROX_SIMPLE);int blackPointsCount = 0;// 在原始图像上绘制轮廓for (int i = 0; i < contours.size(); i++) {Rect rect = opencv_imgproc.boundingRect(contours.get(i));Scalar scalar = new Scalar(0, 255, 0, 0);opencv_imgproc.rectangle(croppedImage, rect, scalar);if (rect.width() > 1 && rect.height() > 1) {blackPointsCount++;}}// 保存标记后的图像opencv_imgcodecs.imwrite("C:\\Users\\HP\\Desktop\\output.jpg", croppedImage);System.out.println("黑点数量： " + blackPointsCount);}
}
对于复杂的图片识别有差距
