java获取iphone手机图片旋转角度问题

问题:手机拍照后，使用微信发送原图给电脑桌面，传到系统，图片会旋转，如下:

该图片2.26M,需要转base64后发送给汉王人脸识别设备,下发用户信息给设备。

但下发的人脸是旋转90度的,没法下发成功。网上找了n多方法,获取原图的长宽是反的,都无法解决，困扰了2 3天。

BufferedImage src = InputImage(srcImgPath);
        if (null != src) {
            int old_w = src.getWidth();
            // 得到源图宽
            int old_h = src.getHeight();

参考原博客地址：https://www.cnblogs.com/zhao1949/p/6473882.html

在maven仓库中引用metadata-extractor-2.3.1

<dependency>
<groupId>com.drewnoakes</groupId>
<artifactId>metadata-extractor</artifactId>
<version>2.10.1</version>
</dependency>

package com.zhx.util.imgutil;

import com.drew.imaging.ImageMetadataReader;
import com.drew.imaging.ImageProcessingException;
import com.drew.imaging.jpeg.JpegProcessingException;
import com.drew.metadata.Directory;
import com.drew.metadata.Metadata;
import com.drew.metadata.MetadataException;
import com.drew.metadata.exif.ExifDirectoryBase;

import javax.imageio.ImageIO;
import java.awt.*;
import java.awt.image.BufferedImage;
import java.io.File;
import java.io.IOException;

;

/**
 * @author SimonHu
 * @Description:
 * @Created on 2018/7/30 22:18
 */
public class RotateImage {

    public static void main(String[] args) throws Exception {
        String filePath = "C:\\Users\\admin\\Desktop\\222222222.jpg";
        String newFilePath ="C:\\Users\\admin\\Desktop\\3333.png";
             RotateImg(filePath,newFilePath);
    }
    public static  boolean RotateImg(String filePath,String newFilePath){
        try {
            File file = new File(filePath);
            //测试发现文件大于7Mb以上时会出现读取速率很慢，找时间再改改；
            Metadata metadata = ImageMetadataReader.readMetadata(file);
            Directory directory = metadata.getFirstDirectoryOfType(ExifDirectoryBase.class);
            int orientation=0;
            // Exif信息中有保存方向,把信息复制到缩略图
            // 原图片的方向信息
            if(directory != null && directory.containsTag(ExifDirectoryBase.TAG_ORIENTATION)){
                orientation = directory.getInt(ExifDirectoryBase.TAG_ORIENTATION);
                System.out.println(orientation);
            }
            int angle=0;
            if(6 == orientation ){
                //6旋转90
                angle = 90;
            }else if( 3 == orientation){
                //3旋转180
                angle = 180;
            }else if( 8 == orientation){
                //8旋转90
                angle = 270;
            }
            BufferedImage src = ImageIO.read(file);
            BufferedImage des = RotateImage.Rotate(src, angle);
            String filename = file.getName();
            String ext = filename.substring(filename.lastIndexOf(".") + 1).toLowerCase();
            ImageIO.write(des,ext, new File(newFilePath));
            return true;
        } catch (JpegProcessingException e) {
            e.printStackTrace();
        } catch (IOException e) {
            e.printStackTrace();
        } catch (MetadataException e) {
            e.printStackTrace();
        } catch (ImageProcessingException e) {
            e.printStackTrace();
        }
        return false;
    }


    public static BufferedImage Rotate(Image src, int angel) {
        int src_width = src.getWidth(null);
        int src_height = src.getHeight(null);
        // calculate the new image size
        Rectangle rect_des = CalcRotatedSize(new Rectangle(new Dimension(
                src_width, src_height)), angel);

        BufferedImage res = null;
        res = new BufferedImage(rect_des.width, rect_des.height,
                BufferedImage.TYPE_INT_RGB);
        Graphics2D g2 = res.createGraphics();
        // transform
        g2.translate((rect_des.width - src_width) / 2,
                (rect_des.height - src_height) / 2);
        g2.rotate(Math.toRadians(angel), src_width / 2, src_height / 2);

        g2.drawImage(src, null, null);
        return res;
    }

    public static Rectangle CalcRotatedSize(Rectangle src, int angel) {
        // if angel is greater than 90 degree, we need to do some conversion
        if (angel >= 90) {
            if(angel / 90 % 2 == 1){
                int temp = src.height;
                src.height = src.width;
                src.width = temp;
            }
            angel = angel % 90;
        }

        double r = Math.sqrt(src.height * src.height + src.width * src.width) / 2;
        double len = 2 * Math.sin(Math.toRadians(angel) / 2) * r;
        double angel_alpha = (Math.PI - Math.toRadians(angel)) / 2;
        double angel_dalta_width = Math.atan((double) src.height / src.width);
        double angel_dalta_height = Math.atan((double) src.width / src.height);

        int len_dalta_width = (int) (len * Math.cos(Math.PI - angel_alpha
                - angel_dalta_width));
        int len_dalta_height = (int) (len * Math.cos(Math.PI - angel_alpha
                - angel_dalta_height));
        int des_width = src.width + len_dalta_width * 2;
        int des_height = src.height + len_dalta_height * 2;
        return new Rectangle(new Dimension(des_width, des_height));
    }

}

在针对微信转发的图片，因为微信的压缩处理，导致图片源信息丢失，JpegMetadataReader.readMetadata或者ImageMetadataReader.readMetadata方法直接处理时会出现“java.lang.NegativeArraySizeException”,这里需要在业务逻辑中进行异常处理。

如此处理后,图片信息不会相反，图片不回旋转，可正常下发到汉王设备,进行人脸识别。