人工智能，百度AI人脸识别java版

人工智能，百度AI人脸识别java版

需求：人脸识别登录，人脸就需要有人脸的照片，数据库建一个字段face保存用户人脸的照片，jquery.webcam.js实现调用摄像头拍照，然后后端接受base64图片，然后与数据库的图片对比，用百度的人脸识别api对比，得出一个分数，分数大于80则登录成功。

前端代码：

<div class="login-desk" id="loginDesk3">
				<div>
					<div id="status">
						<!-- 倒计时 -->
					</div>
					<div id="webcam"></div>
				</div>
				<span id="err_face"></span><br>
				<button id="my-button" type="button" name="login" class="play">登录</button>
			</div>

js代码：

<script src="${pageContext.request.contextPath }/js/webcam/jquery.webcam.min.js"></script> 


<script>
    var w = 320, h = 240;                                      
    var pos = 0, ctx = null, saveCB, image = [];
    var canvas = document.createElement("canvas");
    canvas.setAttribute('width', w);
    canvas.setAttribute('height', h);
    ctx = canvas.getContext("2d");
    image = ctx.getImageData(0, 0, w, h);

    $("#webcam").webcam({
        width: w,
        height: h,
        mode: "callback",                       
        swffile: "${pageContext.request.contextPath }/js/webcam/jscam_canvas_only.swf",
        onTick: function(remain) { 
            if (0 == remain) {
                $("#status").text("登陆中...");
            } else {
                $("#status").text(" "+remain + "秒钟后登录...");
            }
        },
        onSave: function(data){              
            var col = data.split(";");
            var img = image;
            for(var i = 0; i < w; i++) {
                var tmp = parseInt(col[i]);
                img.data[pos + 0] = (tmp >> 16) & 0xff;
                img.data[pos + 1] = (tmp >> 8) & 0xff;
                img.data[pos + 2] = tmp & 0xff;
                img.data[pos + 3] = 0xff;
                pos+= 4;
            }
            if (pos >= 4 * w * h) {
                ctx.putImageData(img,0,0);      
                pos = 0;
                Imagedata=canvas.toDataURL().substring(22);  
                
                $.post("./testFaceLogin3", {type: "data", image: canvas.toDataURL("image/png")}, function(msg){
                    console.log(msg);
                    if(msg==1){
                    	location.href="./user_per/index";
                    }else{
                    	location.href="./login";
                    }
                });
                
            }
        },
        onCapture: function () {               
            webcam.save();      
        },
        debug: function (type, string) {       
            console.log(type + ": " + string);
        },
        onLoad: function() {                  
            console.log('加载完毕！')
            var cams = webcam.getCameraList();
            for(var i in cams) {
            }
        }
    });  

    $(".play").click(function(){
        webcam.capture(5);
    });

</script> 

后端java代码：

//默认传入的参数带类型等参数：data:image/png;base64,
		String imgStr = req.getParameter("image");
		if (null != imgStr) {
			imgStr = imgStr.substring(imgStr.indexOf(",") + 1);
		}
		Boolean flag = GenerateImage(imgStr, filePath, fileName);
		if (flag) {
			
			String image = "img/" + "faceLogin" + "/" + fileName;
			boolean faceFlag = userService.testFaceLogin(image, session, req);
			if (faceFlag) {
				return "1";
			}else{
				session.setAttribute("login_error", "登录失败或者没有开启人脸识别功能");
				return "0";
			}
		}else{
				session.setAttribute("login_error", "登录失败或者没有开启人脸识别功能");
				return "0";
		}

缺点：1、谷歌浏览器开启摄像头需要https支持，IE浏览器摄像头开启无效，小编用的是360极速浏览器。

2、需要安装最新flash

3、浏览器需开通摄像头权限，不能是灰色的。

识别率很高，把自己的证件照上传，然后用有摄登录都识别出来了。

核心算法代码：

/**
	 * true 对比成功
	 * @return
	 */
    public static Boolean match(String img1, String img2) {
        // 请求url
    	Boolean flag = false;
        String url = "https://aip.baidubce.com/rest/2.0/face/v3/match";
        try {
            byte[] bytes1 = FileUtil.readFileByBytes(img1);
            byte[] bytes2 = FileUtil.readFileByBytes(img2);
            String image1 = Base64Util.encode(bytes1);
            String image2 = Base64Util.encode(bytes2);
            List<Map<String, Object>> images = new ArrayList<>();
            Map<String, Object> map1 = new HashMap<>();
            map1.put("image", image1);
            map1.put("image_type", "BASE64");
            map1.put("face_type", "LIVE");
            map1.put("quality_control", "LOW");
            map1.put("liveness_control", "NORMAL");
            Map<String, Object> map2 = new HashMap<>();
            map2.put("image", image2);
            map2.put("image_type", "BASE64");
            map2.put("face_type", "LIVE");
            map2.put("quality_control", "LOW");
            map2.put("liveness_control", "NORMAL");
            images.add(map1);
            images.add(map2);
            String param = GsonUtils.toJson(images);
            // 注意这里仅为了简化编码每一次请求都去获取access_token，线上环境access_token有过期时间， 客户端可自行缓存，过期后重新获取。
            String accessToken = AuthService.getAuth();
            String result = HttpUtil.post(url, accessToken, "application/json", param);
            JSONObject resultJson = parseJSONP(result);
            JSONObject scoreJson = parseJSONP(resultJson.getString("result"));
            String score = scoreJson.getString("score");
            Double obj1 = new Double(score);
            Double obj2 = new Double(80);
            int retval =  obj1.compareTo(obj2);
            if(retval > 0) {
               flag = true;
            }
            else if(retval < 0) {
               flag = false;
            }
            else {
               flag = true;
            }
            //score	是	float	人脸相似度得分，推荐阈值80分
            //{"error_code":0,"error_msg":"SUCCESS","log_id":747956922753388261,"timestamp":1552275338,"cached":0,"result":{"score":12.31431961,"face_list":[{"face_token":"e7c43e08d768787d7a3d5dac8af322f8"},{"face_token":"f7cd6a55d6b61bf643f3417d0b98cfb5"}]}}
            return flag;
        } catch (Exception e) {
            e.printStackTrace();
        }
        return flag;
    }

系统需要整合人脸登录java版的，请加QQ490647751（需付费）。
需要源码学习的，请加QQ490647751（需付费）。