修改相似度算法
parent
b6055e3cbb
commit
ccd0e6cd49
|
|
@ -1,253 +1,454 @@
|
||||||
package com.yanzhu.common.core.utils;
|
package com.yanzhu.common.core.utils;
|
||||||
|
|
||||||
import org.slf4j.Logger;
|
import java.awt.image.BufferedImage;
|
||||||
import org.slf4j.LoggerFactory;
|
import java.io.ByteArrayOutputStream;
|
||||||
import org.opencv.core.*;
|
import java.io.File;
|
||||||
|
import java.io.IOException;
|
||||||
|
import java.io.InputStream;
|
||||||
|
import java.net.URL;
|
||||||
|
import java.net.URLConnection;
|
||||||
|
import java.util.Base64;
|
||||||
|
import javax.imageio.ImageIO;
|
||||||
|
import org.opencv.core.Core;
|
||||||
|
import org.opencv.core.Mat;
|
||||||
|
import org.opencv.core.MatOfRect;
|
||||||
|
import org.opencv.core.Rect;
|
||||||
|
import org.opencv.core.Size;
|
||||||
import org.opencv.imgcodecs.Imgcodecs;
|
import org.opencv.imgcodecs.Imgcodecs;
|
||||||
import org.opencv.imgproc.Imgproc;
|
import org.opencv.imgproc.Imgproc;
|
||||||
import org.opencv.objdetect.CascadeClassifier;
|
import org.opencv.objdetect.CascadeClassifier;
|
||||||
|
|
||||||
import java.io.File;
|
|
||||||
import java.net.URL;
|
|
||||||
import java.nio.file.Files;
|
|
||||||
import java.nio.file.Path;
|
|
||||||
import java.nio.file.StandardCopyOption;
|
|
||||||
|
|
||||||
/**
|
/**
|
||||||
* 图片相似度计算工具类（基于OpenCV实现）
|
* 人像照片相似度算法工具类
|
||||||
* 专门针对人脸识别场景优化的相似度计算
|
* 用于计算两张人脸图片的相似度，支持URL图片和Base64编码图片
|
||||||
*
|
|
||||||
* @author yanzhu
|
|
||||||
* @date 2025-09-03
|
|
||||||
*/
|
*/
|
||||||
public class ImageSimilarityUtils {
|
public class ImageSimilarityUtils {
|
||||||
private static final Logger log = LoggerFactory.getLogger(ImageSimilarityUtils.class);
|
// 相似度阈值，大于此值认为是同一个人
|
||||||
|
private static final double SIMILARITY_THRESHOLD = 0.75;
|
||||||
|
|
||||||
// 人脸识别场景下的考勤阈值（针对同一个人的不同照片）
|
// OpenCV人脸检测器
|
||||||
private static final double ATTENDANCE_THRESHOLD = 0.70;
|
|
||||||
|
|
||||||
// OpenCV相关
|
|
||||||
private static CascadeClassifier faceDetector;
|
private static CascadeClassifier faceDetector;
|
||||||
|
|
||||||
static {
|
static {
|
||||||
try {
|
try {
|
||||||
// 加载OpenCV库
|
// 加载OpenCV库（使用openpnp的OpenCV依赖）
|
||||||
nu.pattern.OpenCV.loadLocally();
|
nu.pattern.OpenCV.loadLocally();
|
||||||
|
// 初始化人脸检测器（使用相对路径）
|
||||||
// 初始化人脸检测器
|
String cascadePath = ImageSimilarityUtils.class.getClassLoader().getResource("opencv/haarcascade_frontalface_default.xml").getPath();
|
||||||
// 注意：在实际部署时，需要确保haarcascade_frontalface_alt.xml文件在classpath中
|
|
||||||
String cascadePath = "haarcascade_frontalface_alt.xml";
|
|
||||||
faceDetector = new CascadeClassifier(cascadePath);
|
faceDetector = new CascadeClassifier(cascadePath);
|
||||||
} catch (Exception e) {
|
} catch (Exception e) {
|
||||||
log.error("OpenCV初始化失败", e);
|
System.err.println("OpenCV加载失败，将使用基础相似度计算方法: " + e.getMessage());
|
||||||
|
faceDetector = null;
|
||||||
}
|
}
|
||||||
}
|
}
|
||||||
|
|
||||||
/**
|
/**
|
||||||
* 计算两张人脸图片的相似度
|
* 计算两张人脸图片的相似度
|
||||||
*
|
*
|
||||||
* @param imagePath1 人脸图片1路径（可以是本地路径或URL）
|
* @param imageUrl1 第一张图片URL
|
||||||
* @param imagePath2 人脸图片2路径（可以是本地路径或URL）
|
* @param imageUrl2 第二张图片URL
|
||||||
* @return 相似度（0-1之间，1表示完全相同）
|
* @return 相似度，范围0-1，越大表示越相似
|
||||||
*/
|
*/
|
||||||
public static double calculateFaceSimilarity(String imagePath1, String imagePath2) {
|
public static double calculateFaceSimilarity(String imageUrl1, String imageUrl2) {
|
||||||
try {
|
try {
|
||||||
// 检查参数
|
// 下载图片
|
||||||
if (imagePath1 == null || imagePath2 == null) {
|
BufferedImage image1 = downloadImage(imageUrl1);
|
||||||
log.warn("图片路径不能为空");
|
BufferedImage image2 = downloadImage(imageUrl2);
|
||||||
|
|
||||||
|
if (image1 == null || image2 == null) {
|
||||||
return 0.0;
|
return 0.0;
|
||||||
}
|
}
|
||||||
|
|
||||||
// 读取图片
|
// 提取特征并计算相似度
|
||||||
Mat image1 = readImageAsMat(imagePath1);
|
return calculateImageSimilarity(image1, image2);
|
||||||
Mat image2 = readImageAsMat(imagePath2);
|
|
||||||
|
|
||||||
if (image1.empty() || image2.empty()) {
|
|
||||||
log.warn("无法读取图片文件: {} 或 {}", imagePath1, imagePath2);
|
|
||||||
return 0.0;
|
|
||||||
}
|
|
||||||
|
|
||||||
// 检测并裁剪人脸区域
|
|
||||||
Mat face1 = detectAndCropFace(image1);
|
|
||||||
Mat face2 = detectAndCropFace(image2);
|
|
||||||
|
|
||||||
// 如果无法检测到人脸，则使用整张图片
|
|
||||||
if (face1.empty()) face1 = image1;
|
|
||||||
if (face2.empty()) face2 = image2;
|
|
||||||
|
|
||||||
// 为人脸识别场景预处理
|
|
||||||
Mat processedImage1 = preprocessForFaceRecognition(face1);
|
|
||||||
Mat processedImage2 = preprocessForFaceRecognition(face2);
|
|
||||||
|
|
||||||
// 使用OpenCV计算直方图相似度
|
|
||||||
double similarity = calculateHistogramSimilarity(processedImage1, processedImage2);
|
|
||||||
|
|
||||||
log.debug("计算人脸图片相似度完成: {} 和 {}, 相似度: {}",
|
|
||||||
imagePath1, imagePath2, similarity);
|
|
||||||
|
|
||||||
// 释放资源
|
|
||||||
image1.release();
|
|
||||||
image2.release();
|
|
||||||
face1.release();
|
|
||||||
face2.release();
|
|
||||||
processedImage1.release();
|
|
||||||
processedImage2.release();
|
|
||||||
|
|
||||||
return similarity;
|
|
||||||
} catch (Exception e) {
|
} catch (Exception e) {
|
||||||
log.error("计算人脸图片相似度时发生错误", e);
|
e.printStackTrace();
|
||||||
return 0.0;
|
return 0.0;
|
||||||
}
|
}
|
||||||
}
|
}
|
||||||
|
|
||||||
/**
|
/**
|
||||||
* 检查人脸图片是否匹配（基于考勤场景的阈值）
|
* 计算两张Base64编码人脸图片的相似度
|
||||||
*
|
*
|
||||||
* @param imagePath1 人脸图片1路径
|
* @param imageBase64_1 第一张图片的Base64编码
|
||||||
* @param imagePath2 人脸图片2路径
|
* @param imageBase64_2 第二张图片的Base64编码
|
||||||
* @return 是否匹配
|
* @return 相似度，范围0-1，越大表示越相似
|
||||||
*/
|
*/
|
||||||
public static boolean isFaceMatchForAttendance(String imagePath1, String imagePath2) {
|
public static double calculateFaceSimilarityByBase64(String imageBase64_1, String imageBase64_2) {
|
||||||
double similarity = calculateFaceSimilarity(imagePath1, imagePath2);
|
|
||||||
return similarity >= ATTENDANCE_THRESHOLD;
|
|
||||||
}
|
|
||||||
|
|
||||||
/**
|
|
||||||
* 读取图片为OpenCV的Mat对象
|
|
||||||
*
|
|
||||||
* @param imagePath 图片路径
|
|
||||||
* @return Mat对象
|
|
||||||
*/
|
|
||||||
private static Mat readImageAsMat(String imagePath) {
|
|
||||||
try {
|
try {
|
||||||
if (imagePath.startsWith("http://") || imagePath.startsWith("https://")) {
|
// 解码Base64图片
|
||||||
// 处理网络图片
|
BufferedImage image1 = decodeBase64Image(imageBase64_1);
|
||||||
URL url = new URL(imagePath);
|
BufferedImage image2 = decodeBase64Image(imageBase64_2);
|
||||||
Path tempFile = Files.createTempFile("image_", ".jpg");
|
|
||||||
Files.copy(url.openStream(), tempFile, StandardCopyOption.REPLACE_EXISTING);
|
if (image1 == null || image2 == null) {
|
||||||
Mat mat = Imgcodecs.imread(tempFile.toAbsolutePath().toString());
|
return 0.0;
|
||||||
Files.delete(tempFile); // 清理临时文件
|
|
||||||
return mat;
|
|
||||||
} else {
|
|
||||||
// 处理本地图片
|
|
||||||
return Imgcodecs.imread(imagePath);
|
|
||||||
}
|
}
|
||||||
|
|
||||||
|
// 提取特征并计算相似度
|
||||||
|
return calculateImageSimilarity(image1, image2);
|
||||||
} catch (Exception e) {
|
} catch (Exception e) {
|
||||||
log.error("读取图片失败: {}", imagePath, e);
|
e.printStackTrace();
|
||||||
return new Mat();
|
return 0.0;
|
||||||
}
|
}
|
||||||
}
|
}
|
||||||
|
|
||||||
/**
|
/**
|
||||||
* 检测并裁剪人脸区域
|
* 判断两张人脸图片是否匹配（用于考勤等场景）
|
||||||
*
|
*
|
||||||
* @param image 原始图片
|
* @param imageUrl1 第一张图片URL
|
||||||
* @return 裁剪后的人脸区域图片，如果未检测到则返回空Mat
|
* @param imageUrl2 第二张图片URL
|
||||||
|
* @return 是否匹配
|
||||||
*/
|
*/
|
||||||
private static Mat detectAndCropFace(Mat image) {
|
public static boolean isFaceMatchForAttendance(String imageUrl1, String imageUrl2) {
|
||||||
|
double similarity = calculateFaceSimilarity(imageUrl1, imageUrl2);
|
||||||
|
return similarity >= SIMILARITY_THRESHOLD;
|
||||||
|
}
|
||||||
|
|
||||||
|
/**
|
||||||
|
* 判断两张Base64编码人脸图片是否匹配（用于考勤等场景）
|
||||||
|
*
|
||||||
|
* @param imageBase64_1 第一张图片的Base64编码
|
||||||
|
* @param imageBase64_2 第二张图片的Base64编码
|
||||||
|
* @return 是否匹配
|
||||||
|
*/
|
||||||
|
public static boolean isFaceMatchForAttendanceByBase64(String imageBase64_1, String imageBase64_2) {
|
||||||
|
double similarity = calculateFaceSimilarityByBase64(imageBase64_1, imageBase64_2);
|
||||||
|
return similarity >= SIMILARITY_THRESHOLD;
|
||||||
|
}
|
||||||
|
|
||||||
|
/**
|
||||||
|
* 下载图片
|
||||||
|
*
|
||||||
|
* @param imageUrl 图片URL
|
||||||
|
* @return BufferedImage对象
|
||||||
|
*/
|
||||||
|
private static BufferedImage downloadImage(String imageUrl) {
|
||||||
try {
|
try {
|
||||||
|
URL url = new URL(imageUrl);
|
||||||
|
URLConnection connection = url.openConnection();
|
||||||
|
connection.setConnectTimeout(5000);
|
||||||
|
connection.setReadTimeout(5000);
|
||||||
|
return ImageIO.read(connection.getInputStream());
|
||||||
|
} catch (Exception e) {
|
||||||
|
e.printStackTrace();
|
||||||
|
return null;
|
||||||
|
}
|
||||||
|
}
|
||||||
|
|
||||||
|
/**
|
||||||
|
* 解码Base64图片
|
||||||
|
*
|
||||||
|
* @param base64Image Base64编码的图片字符串
|
||||||
|
* @return BufferedImage对象
|
||||||
|
* @throws IOException 解码异常
|
||||||
|
*/
|
||||||
|
private static BufferedImage decodeBase64Image(String base64Image) throws IOException {
|
||||||
|
// 移除可能存在的Base64前缀（如data:image/jpeg;base64,）
|
||||||
|
if (base64Image.contains(",")) {
|
||||||
|
base64Image = base64Image.split(",")[1];
|
||||||
|
}
|
||||||
|
|
||||||
|
byte[] imageBytes = Base64.getDecoder().decode(base64Image);
|
||||||
|
return ImageIO.read(new java.io.ByteArrayInputStream(imageBytes));
|
||||||
|
}
|
||||||
|
|
||||||
|
/**
|
||||||
|
* 将图片转换为Base64编码
|
||||||
|
*
|
||||||
|
* @param image 图片对象
|
||||||
|
* @param formatName 图片格式（如"jpg"、"png"）
|
||||||
|
* @return Base64编码字符串
|
||||||
|
* @throws IOException 编码异常
|
||||||
|
*/
|
||||||
|
private static String encodeImageToBase64(BufferedImage image, String formatName) throws IOException {
|
||||||
|
ByteArrayOutputStream baos = new ByteArrayOutputStream();
|
||||||
|
ImageIO.write(image, formatName, baos);
|
||||||
|
byte[] imageBytes = baos.toByteArray();
|
||||||
|
return Base64.getEncoder().encodeToString(imageBytes);
|
||||||
|
}
|
||||||
|
|
||||||
|
/**
|
||||||
|
* 计算两张图片的相似度
|
||||||
|
* 使用OpenCV进行人脸检测，然后比较人脸区域的直方图
|
||||||
|
*
|
||||||
|
* @param image1 第一张图片
|
||||||
|
* @param image2 第二张图片
|
||||||
|
* @return 相似度，范围0-1
|
||||||
|
*/
|
||||||
|
private static double calculateImageSimilarity(BufferedImage image1, BufferedImage image2) {
|
||||||
|
// 如果OpenCV初始化失败，直接使用基础方法
|
||||||
|
if (faceDetector == null) {
|
||||||
|
return calculateImageSimilarityBasic(image1, image2);
|
||||||
|
}
|
||||||
|
|
||||||
|
try {
|
||||||
|
// 检测人脸区域
|
||||||
|
Rect face1 = detectFace(image1);
|
||||||
|
Rect face2 = detectFace(image2);
|
||||||
|
|
||||||
|
// 如果检测到人脸，使用人脸区域进行比较
|
||||||
|
if (face1 != null && face2 != null) {
|
||||||
|
BufferedImage faceImage1 = cropFace(image1, face1);
|
||||||
|
BufferedImage faceImage2 = cropFace(image2, face2);
|
||||||
|
|
||||||
|
// 将人脸区域缩放到相同大小
|
||||||
|
int width = 100;
|
||||||
|
int height = 100;
|
||||||
|
BufferedImage scaledFace1 = scaleImage(faceImage1, width, height);
|
||||||
|
BufferedImage scaledFace2 = scaleImage(faceImage2, width, height);
|
||||||
|
|
||||||
|
// 计算人脸区域的灰度直方图
|
||||||
|
int[] histogram1 = calculateHistogram(scaledFace1);
|
||||||
|
int[] histogram2 = calculateHistogram(scaledFace2);
|
||||||
|
|
||||||
|
// 计算直方图相似度（余弦相似度）
|
||||||
|
return calculateCosineSimilarity(histogram1, histogram2);
|
||||||
|
}
|
||||||
|
|
||||||
|
// 如果未检测到人脸，使用整个图片进行比较
|
||||||
|
int width = 100;
|
||||||
|
int height = 100;
|
||||||
|
BufferedImage scaledImage1 = scaleImage(image1, width, height);
|
||||||
|
BufferedImage scaledImage2 = scaleImage(image2, width, height);
|
||||||
|
|
||||||
|
// 计算图片的灰度直方图
|
||||||
|
int[] histogram1 = calculateHistogram(scaledImage1);
|
||||||
|
int[] histogram2 = calculateHistogram(scaledImage2);
|
||||||
|
|
||||||
|
// 计算直方图相似度（余弦相似度）
|
||||||
|
return calculateCosineSimilarity(histogram1, histogram2);
|
||||||
|
|
||||||
|
} catch (Exception e) {
|
||||||
|
e.printStackTrace();
|
||||||
|
// 发生异常时使用原始方法
|
||||||
|
return calculateImageSimilarityBasic(image1, image2);
|
||||||
|
}
|
||||||
|
}
|
||||||
|
|
||||||
|
/**
|
||||||
|
* 基础图片相似度计算方法（不使用OpenCV）
|
||||||
|
* 使用直方图比较方法计算相似度
|
||||||
|
*
|
||||||
|
* @param image1 第一张图片
|
||||||
|
* @param image2 第二张图片
|
||||||
|
* @return 相似度，范围0-1
|
||||||
|
*/
|
||||||
|
private static double calculateImageSimilarityBasic(BufferedImage image1, BufferedImage image2) {
|
||||||
|
// 将图片缩放到相同大小
|
||||||
|
int width = 100;
|
||||||
|
int height = 100;
|
||||||
|
BufferedImage scaledImage1 = scaleImage(image1, width, height);
|
||||||
|
BufferedImage scaledImage2 = scaleImage(image2, width, height);
|
||||||
|
|
||||||
|
// 计算图片的灰度直方图
|
||||||
|
int[] histogram1 = calculateHistogram(scaledImage1);
|
||||||
|
int[] histogram2 = calculateHistogram(scaledImage2);
|
||||||
|
|
||||||
|
// 计算直方图相似度（余弦相似度）
|
||||||
|
return calculateCosineSimilarity(histogram1, histogram2);
|
||||||
|
}
|
||||||
|
|
||||||
|
/**
|
||||||
|
* 检测图片中的人脸
|
||||||
|
*
|
||||||
|
* @param image 输入图片
|
||||||
|
* @return 人脸区域矩形，如果未检测到人脸则返回null
|
||||||
|
*/
|
||||||
|
private static Rect detectFace(BufferedImage image) {
|
||||||
|
try {
|
||||||
|
// 将BufferedImage转换为OpenCV Mat
|
||||||
|
Mat mat = bufferedImageToMat(image);
|
||||||
|
|
||||||
// 转换为灰度图
|
// 转换为灰度图
|
||||||
Mat grayImage = new Mat();
|
Mat gray = new Mat();
|
||||||
Imgproc.cvtColor(image, grayImage, Imgproc.COLOR_BGR2GRAY);
|
Imgproc.cvtColor(mat, gray, Imgproc.COLOR_BGR2GRAY);
|
||||||
|
|
||||||
// 检测人脸
|
// 检测人脸
|
||||||
MatOfRect faceDetections = new MatOfRect();
|
MatOfRect faceDetections = new MatOfRect();
|
||||||
faceDetector.detectMultiScale(grayImage, faceDetections);
|
faceDetector.detectMultiScale(gray, faceDetections, 1.1, 3, 0, new Size(30, 30), new Size());
|
||||||
|
|
||||||
Rect[] faces = faceDetections.toArray();
|
Rect[] faces = faceDetections.toArray();
|
||||||
|
|
||||||
// 如果检测到人脸，裁剪第一个人脸区域
|
|
||||||
if (faces.length > 0) {
|
if (faces.length > 0) {
|
||||||
Rect face = faces[0];
|
// 返回最大的人脸区域
|
||||||
Mat croppedFace = new Mat(image, face);
|
return faces[0];
|
||||||
grayImage.release();
|
|
||||||
faceDetections.release();
|
|
||||||
return croppedFace;
|
|
||||||
}
|
}
|
||||||
|
|
||||||
grayImage.release();
|
return null;
|
||||||
faceDetections.release();
|
|
||||||
return new Mat();
|
|
||||||
} catch (Exception e) {
|
} catch (Exception e) {
|
||||||
log.warn("人脸检测失败，使用整张图片", e);
|
e.printStackTrace();
|
||||||
|
return null;
|
||||||
|
}
|
||||||
|
}
|
||||||
|
|
||||||
|
/**
|
||||||
|
* 裁剪人脸区域
|
||||||
|
*
|
||||||
|
* @param image 原图片
|
||||||
|
* @param faceRect 人脸区域
|
||||||
|
* @return 裁剪后的人脸图片
|
||||||
|
*/
|
||||||
|
private static BufferedImage cropFace(BufferedImage image, Rect faceRect) {
|
||||||
|
// 确保人脸区域在图片范围内
|
||||||
|
int x = Math.max(0, faceRect.x);
|
||||||
|
int y = Math.max(0, faceRect.y);
|
||||||
|
int width = Math.min(image.getWidth() - x, faceRect.width);
|
||||||
|
int height = Math.min(image.getHeight() - y, faceRect.height);
|
||||||
|
|
||||||
|
return image.getSubimage(x, y, width, height);
|
||||||
|
}
|
||||||
|
|
||||||
|
/**
|
||||||
|
* 将BufferedImage转换为OpenCV Mat
|
||||||
|
*
|
||||||
|
* @param image BufferedImage对象
|
||||||
|
* @return OpenCV Mat对象
|
||||||
|
*/
|
||||||
|
private static Mat bufferedImageToMat(BufferedImage image) {
|
||||||
|
// 创建临时文件
|
||||||
|
try {
|
||||||
|
File tempFile = File.createTempFile("opencv", ".jpg");
|
||||||
|
ImageIO.write(image, "jpg", tempFile);
|
||||||
|
|
||||||
|
// 读取为Mat
|
||||||
|
Mat mat = Imgcodecs.imread(tempFile.getAbsolutePath());
|
||||||
|
|
||||||
|
// 删除临时文件
|
||||||
|
tempFile.delete();
|
||||||
|
|
||||||
|
return mat;
|
||||||
|
} catch (Exception e) {
|
||||||
|
e.printStackTrace();
|
||||||
return new Mat();
|
return new Mat();
|
||||||
}
|
}
|
||||||
}
|
}
|
||||||
|
|
||||||
/**
|
/**
|
||||||
* 为人脸识别场景预处理图片
|
* 缩放图片到指定大小
|
||||||
*
|
*
|
||||||
* @param originalImage 原始图片
|
* @param image 原图片
|
||||||
* @return 预处理后的图片
|
* @param width 目标宽度
|
||||||
|
* @param height 目标高度
|
||||||
|
* @return 缩放后的图片
|
||||||
*/
|
*/
|
||||||
private static Mat preprocessForFaceRecognition(Mat originalImage) {
|
private static BufferedImage scaleImage(BufferedImage image, int width, int height) {
|
||||||
// 1. 调整大小到标准尺寸（256x256）
|
BufferedImage scaledImage = new BufferedImage(width, height, BufferedImage.TYPE_INT_RGB);
|
||||||
Mat resized = new Mat();
|
scaledImage.getGraphics().drawImage(image, 0, 0, width, height, null);
|
||||||
Imgproc.resize(originalImage, resized, new Size(256, 256));
|
return scaledImage;
|
||||||
|
|
||||||
// 2. 转换为灰度图（如果还不是的话）
|
|
||||||
Mat gray = new Mat();
|
|
||||||
if (resized.channels() == 3) {
|
|
||||||
Imgproc.cvtColor(resized, gray, Imgproc.COLOR_BGR2GRAY);
|
|
||||||
} else {
|
|
||||||
resized.copyTo(gray);
|
|
||||||
}
|
|
||||||
|
|
||||||
// 3. 直方图均衡化增强对比度
|
|
||||||
Mat enhanced = new Mat();
|
|
||||||
Imgproc.equalizeHist(gray, enhanced);
|
|
||||||
|
|
||||||
// 释放中间资源
|
|
||||||
resized.release();
|
|
||||||
gray.release();
|
|
||||||
|
|
||||||
return enhanced;
|
|
||||||
}
|
}
|
||||||
|
|
||||||
/**
|
/**
|
||||||
* 计算直方图相似度（使用相关性比较）
|
* 计算图片的灰度直方图
|
||||||
*
|
*
|
||||||
* @param image1 图片1
|
* @param image 图片
|
||||||
* @param image2 图片2
|
* @return 灰度直方图（256个灰度级）
|
||||||
* @return 相似度（0-1之间）
|
|
||||||
*/
|
*/
|
||||||
private static double calculateHistogramSimilarity(Mat image1, Mat image2) {
|
private static int[] calculateHistogram(BufferedImage image) {
|
||||||
// 计算直方图
|
int[] histogram = new int[256];
|
||||||
Mat hist1 = new Mat();
|
int width = image.getWidth();
|
||||||
Mat hist2 = new Mat();
|
int height = image.getHeight();
|
||||||
|
|
||||||
// 设置直方图参数
|
for (int y = 0; y < height; y++) {
|
||||||
MatOfInt histSize = new MatOfInt(256);
|
for (int x = 0; x < width; x++) {
|
||||||
MatOfFloat ranges = new MatOfFloat(0f, 256f);
|
int rgb = image.getRGB(x, y);
|
||||||
MatOfInt channels = new MatOfInt(0);
|
int r = (rgb >> 16) & 0xFF;
|
||||||
|
int g = (rgb >> 8) & 0xFF;
|
||||||
|
int b = rgb & 0xFF;
|
||||||
|
|
||||||
// 计算直方图
|
// 转换为灰度值 (0.299*R + 0.587*G + 0.114*B)
|
||||||
Imgproc.calcHist(java.util.Arrays.asList(image1), channels, new Mat(), hist1, histSize, ranges);
|
int gray = (int)(0.299 * r + 0.587 * g + 0.114 * b);
|
||||||
Imgproc.calcHist(java.util.Arrays.asList(image2), channels, new Mat(), hist2, histSize, ranges);
|
histogram[gray]++;
|
||||||
|
}
|
||||||
|
}
|
||||||
|
|
||||||
// 归一化直方图
|
return histogram;
|
||||||
Core.normalize(hist1, hist1);
|
}
|
||||||
Core.normalize(hist2, hist2);
|
|
||||||
|
|
||||||
// 计算直方图相关性
|
/**
|
||||||
double similarity = Imgproc.compareHist(hist1, hist2, Imgproc.HISTCMP_CORREL);
|
* 计算余弦相似度
|
||||||
|
*
|
||||||
|
* @param vector1 向量1
|
||||||
|
* @param vector2 向量2
|
||||||
|
* @return 余弦相似度，范围0-1
|
||||||
|
*/
|
||||||
|
private static double calculateCosineSimilarity(int[] vector1, int[] vector2) {
|
||||||
|
double dotProduct = 0.0;
|
||||||
|
double norm1 = 0.0;
|
||||||
|
double norm2 = 0.0;
|
||||||
|
|
||||||
// 释放资源
|
for (int i = 0; i < vector1.length; i++) {
|
||||||
hist1.release();
|
dotProduct += vector1[i] * vector2[i];
|
||||||
hist2.release();
|
norm1 += Math.pow(vector1[i], 2);
|
||||||
histSize.release();
|
norm2 += Math.pow(vector2[i], 2);
|
||||||
ranges.release();
|
}
|
||||||
channels.release();
|
|
||||||
|
|
||||||
// 确保相似度在0-1范围内
|
// 避免除以零
|
||||||
return Math.max(0.0, Math.min(1.0, similarity));
|
if (norm1 == 0 || norm2 == 0) {
|
||||||
|
return 0.0;
|
||||||
|
}
|
||||||
|
|
||||||
|
return dotProduct / (Math.sqrt(norm1) * Math.sqrt(norm2));
|
||||||
}
|
}
|
||||||
|
|
||||||
public static void main(String[] args) {
|
public static void main(String[] args) {
|
||||||
String userPicture = "https://xiangguan.sxyanzhu.com/statics/2025/09/03/836e5c21f83604894486069394dcd22e_20250903170015A004.jpg";
|
System.out.println("开始测试人脸相似度算法...");
|
||||||
String attImg = "https://xiangguan.sxyanzhu.com/statics/2025/09/03/c4e2c8fb9d9f66e03902f9e3fea17f99_20250903165717A003.jpg";
|
|
||||||
double similarity = ImageSimilarityUtils.calculateFaceSimilarity(userPicture, attImg);
|
// 测试1：同一个人的不同照片（应该匹配）
|
||||||
System.out.println("人脸相似度: " + similarity);
|
String userPicture1 = "https://xiangguan.sxyanzhu.com/statics/2025/09/03/836e5c21f83604894486069394dcd22e_20250903170015A004.jpg";
|
||||||
System.out.println("考勤匹配: " + ImageSimilarityUtils.isFaceMatchForAttendance(userPicture, attImg));
|
String userPicture2 = "http://62.234.3.186/statics/2025/06/11/a28457e2847333886c8e0f4fd9cd24bc_20250611101313A331.jpg";
|
||||||
|
String userPicture3 = "https://xiangguan.sxyanzhu.com/statics/2025/09/03/c4e2c8fb9d9f66e03902f9e3fea17f99_20250903165717A003.jpg";
|
||||||
|
String userPicture4 = "https://xiangguan.sxyanzhu.com/statics/2025/09/03/8c7dd922ad47494fc02c388e12c00eac_20250903132353A852.png";
|
||||||
|
String userPicture5 = "http://62.234.3.186/statics/2025/06/11/87052f8fa3eaa8840bc2e4fe556a825e_20250611101011A328.jpg";
|
||||||
|
|
||||||
|
|
||||||
|
String img1=userPicture4;
|
||||||
|
String img2=userPicture5;
|
||||||
|
System.out.println("\n测试1:不同人的不同照片");
|
||||||
|
System.out.println("照片2: " + img1);
|
||||||
|
System.out.println("照片5: " + img2);
|
||||||
|
|
||||||
|
double similarity1 = ImageSimilarityUtils.calculateFaceSimilarity(img1, img2);
|
||||||
|
System.out.println("人脸相似度: " + similarity1);
|
||||||
|
System.out.println("考勤匹配: " + ImageSimilarityUtils.isFaceMatchForAttendance(img1, img2));
|
||||||
|
|
||||||
|
// System.out.println("\n测试2：同一个人的另一组照片");
|
||||||
|
// System.out.println("照片1: " + userPicture1);
|
||||||
|
// System.out.println("照片4: " + userPicture4);
|
||||||
|
|
||||||
|
// double similarity2 = ImageSimilarityUtils.calculateFaceSimilarity(userPicture1, userPicture4);
|
||||||
|
// System.out.println("人脸相似度: " + similarity2);
|
||||||
|
// System.out.println("考勤匹配: " + ImageSimilarityUtils.isFaceMatchForAttendance(userPicture1, userPicture4));
|
||||||
|
|
||||||
|
// System.out.println("\n测试2：同一个人的另一组照片");
|
||||||
|
// System.out.println("照片1: " + userPicture1);
|
||||||
|
// System.out.println("照片2: " + userPicture2);
|
||||||
|
|
||||||
|
// double similarity3 = ImageSimilarityUtils.calculateFaceSimilarity(userPicture1, userPicture2);
|
||||||
|
// System.out.println("人脸相似度: " + similarity3);
|
||||||
|
// System.out.println("考勤匹配: " + ImageSimilarityUtils.isFaceMatchForAttendance(userPicture1, userPicture2));
|
||||||
|
|
||||||
|
// // 测试Base64图片人脸相似度计算
|
||||||
|
// // 实际使用时，这里应该是真实的Base64编码图片字符串
|
||||||
|
// System.out.println("\n测试3：Base64图片相似度测试");
|
||||||
|
// try {
|
||||||
|
// // 将URL图片转为Base64进行测试
|
||||||
|
// BufferedImage img1 = downloadImage(userPicture1);
|
||||||
|
// BufferedImage img2 = downloadImage(userPicture3);
|
||||||
|
// if (img1 != null && img2 != null) {
|
||||||
|
// String base64Image1 = encodeImageToBase64(img1, "jpg");
|
||||||
|
// String base64Image2 = encodeImageToBase64(img2, "jpg");
|
||||||
|
|
||||||
|
// double base64Similarity = calculateFaceSimilarityByBase64(base64Image1, base64Image2);
|
||||||
|
// System.out.println("Base64人脸相似度: " + base64Similarity);
|
||||||
|
|
||||||
|
// boolean base64IsMatch = isFaceMatchForAttendanceByBase64(base64Image1, base64Image2);
|
||||||
|
// System.out.println("Base64是否匹配: " + base64IsMatch);
|
||||||
|
// } else {
|
||||||
|
// System.out.println("无法下载图片进行Base64测试");
|
||||||
|
// }
|
||||||
|
// } catch (Exception e) {
|
||||||
|
// System.out.println("Base64图片测试失败: " + e.getMessage());
|
||||||
|
// }
|
||||||
|
|
||||||
|
System.out.println("\n测试完成。");
|
||||||
}
|
}
|
||||||
}
|
}
|
||||||
File diff suppressed because it is too large
Load Diff
File diff suppressed because it is too large
Load Diff
Loading…
Reference in New Issue