公司项目需要人脸识别登录,需要支持离线识别,所以无法使用在线的人脸识别的API,于是使用到了离线SDK来对比识别人脸相识度.
获取人脸抓拍的图片需要对接设备,这里不做记录,假设我们已经获取到了人脸图片,只调用SDK将抓拍的人脸图片与已有的进行人脸库图片对比,具体步骤如下:
1. 注册账号,实名认证
https://ai.arcsoft.com.cn/ucenter/user/userlogin
2. 创建应用,在我的应用里面添加SDK,下载sdk到本地
3. 解压SDK,将SDK下的libs文件夹的库文件放到jdk下的lib文件夹下
4. 将SDK libs文件夹下的jar文件导入项目,然后buildpath
5. 编写测试代码,进行人脸比对
package net.nblh.faceRecognition.util.faceCompare;
import static com.arcsoft.face.toolkit.ImageFactory.getRGBData;
import java.io.File;
import java.math.BigDecimal;
import java.util.ArrayList;
import java.util.List;
import com.arcsoft.face.EngineConfiguration;
import com.arcsoft.face.FaceEngine;
import com.arcsoft.face.FaceFeature;
import com.arcsoft.face.FaceInfo;
import com.arcsoft.face.FaceSimilar;
import com.arcsoft.face.FunctionConfiguration;
import com.arcsoft.face.enums.DetectMode;
import com.arcsoft.face.enums.DetectOrient;
import com.arcsoft.face.enums.ErrorInfo;
import com.arcsoft.face.enums.ImageFormat;
import com.arcsoft.face.toolkit.ImageInfo;
/**
* 使用虹软SDK 进行人脸比对
* @author qxl
*/
public class ArcFaceCompare {
/**
* 初始化引擎
* @param appId
* @param sdkKey
* @return
*/
public static FaceEngine initFaceEngine(String appId,String sdkKey) {
FaceEngine faceEngine = new FaceEngine();
// 激活引擎
int activeCode = faceEngine.activeOnline(appId, sdkKey);
if (activeCode != ErrorInfo.MOK.getValue() && activeCode != ErrorInfo.MERR_ASF_ALREADY_ACTIVATED.getValue()) {
System.out.println("引擎激活失败");
return null;
}
// 引擎配置
EngineConfiguration engineConfiguration = new EngineConfiguration();
engineConfiguration.setDetectMode(DetectMode.ASF_DETECT_MODE_IMAGE);
engineConfiguration.setDetectFaceOrientPriority(DetectOrient.ASF_OP_0_ONLY);
// 功能配置
FunctionConfiguration functionConfiguration = new FunctionConfiguration();
functionConfiguration.setSupportAge(true);
functionConfiguration.setSupportFace3dAngle(true);
functionConfiguration.setSupportFaceDetect(true);
functionConfiguration.setSupportFaceRecognition(true);
functionConfiguration.setSupportGender(true);
functionConfiguration.setSupportLiveness(true);
functionConfiguration.setSupportIRLiveness(true);
engineConfiguration.setFunctionConfiguration(functionConfiguration);
// 初始化引擎
int initCode = faceEngine.init(engineConfiguration);
if (initCode != ErrorInfo.MOK.getValue()) {
System.out.println("初始化引擎失败");
return null;
}
return faceEngine;
}
/**
* 人脸图片比对
* @param faceEngine 人脸识别引擎对象
* @param picPath1 人脸图片1路径
* @param picPath2 人脸图片2路径
* @return
*/
public static float faceCompare(FaceEngine faceEngine,String picPath1,String picPath2) {
// 人脸检测
ImageInfo imageInfo = getRGBData(new File(picPath1));
List<FaceInfo> faceInfoList = new ArrayList<FaceInfo>();
int detectCode = faceEngine.detectFaces(imageInfo.getImageData(), imageInfo.getWidth(), imageInfo.getHeight(),
ImageFormat.CP_PAF_BGR24, faceInfoList);
//System.out.println(faceInfoList);
// 特征提取
FaceFeature faceFeature = new FaceFeature();
int extractCode = faceEngine.extractFaceFeature(imageInfo.getImageData(), imageInfo.getWidth(),
imageInfo.getHeight(), ImageFormat.CP_PAF_BGR24, faceInfoList.get(0), faceFeature);
//System.out.println("特征值大小：" + faceFeature.getFeatureData().length);
// 人脸检测2
ImageInfo imageInfo2 = getRGBData(new File(picPath2));
List<FaceInfo> faceInfoList2 = new ArrayList<FaceInfo>();
int detectCode2 = faceEngine.detectFaces(imageInfo2.getImageData(), imageInfo2.getWidth(),
imageInfo2.getHeight(), ImageFormat.CP_PAF_BGR24, faceInfoList2);
//System.out.println(faceInfoList);
// 特征提取2
FaceFeature faceFeature2 = new FaceFeature();
int extractCode2 = faceEngine.extractFaceFeature(imageInfo2.getImageData(), imageInfo2.getWidth(),
imageInfo2.getHeight(), ImageFormat.CP_PAF_BGR24, faceInfoList2.get(0), faceFeature2);
//System.out.println("特征值大小：" + faceFeature.getFeatureData().length);
// 特征比对
FaceFeature targetFaceFeature = new FaceFeature();
targetFaceFeature.setFeatureData(faceFeature.getFeatureData());
FaceFeature sourceFaceFeature = new FaceFeature();
sourceFaceFeature.setFeatureData(faceFeature2.getFeatureData());
FaceSimilar faceSimilar = new FaceSimilar();
int compareCode = faceEngine.compareFaceFeature(targetFaceFeature, sourceFaceFeature, faceSimilar);
float score = faceSimilar.getScore();
BigDecimal b = new BigDecimal(score);
//四舍五入,保留四位小数
float similarity =b.setScale(4,BigDecimal.ROUND_HALF_UP).floatValue();
System.out.println("相似度：" + similarity);
return similarity;
}
/**
* 卸载引擎
* @param faceEngine
*/
public static void uninstall(FaceEngine faceEngine) {
faceEngine.unInit();
}
public static void main(String[] args) {
String appId = "图2创建应用的APP ID";
String sdkKey = "图2创建应用的SDK Key";
//初始化引擎
FaceEngine faceEngine = initFaceEngine(appId,sdkKey);
//进行人脸比对
String picPath1 = "F:\\qxlTest\\1.jpg";
String picPath2 = "F:\\qxlTest\\9.jpg";
float similarity = faceCompare(faceEngine, picPath1, picPath2);
//卸载引擎
uninstall(faceEngine);
}
}
6.运行main方法测试
注意事项:
A: 虹软人脸识别离线 SDK 每台机器对应一个SDK,第一次激活SDK会与机器绑定,不同的机器需要下载新的SDK
B: 离线 SDK自下载之日起,有效期一年,一年使用时间过后,需要重新下载最新版,同时更新对应的 dll文件, appid 和sdkkey,更新之后请更新以上的APP ID 和 SDK KEY
C: 具体详见官网 https://ai.arcsoft.com.cn/