Java 提取和删除照片图片 Exif GPS 等隐私信息

照片图片 Exif

通过手机相机或者数码相机拍摄的照片都带有 Exif 元数据信息，比如下面这张照片：

它的 Exif 信息为：

Root: 
	ImageWidth: 4000
	ImageLength: 3000
	Make: 'Xiaomi'
	Model: 'MI CC 9'
	Orientation: 1
	XResolution: 72
	YResolution: 72
	ResolutionUnit: 2
	DateTime: '2019:12:16 13:49:17'
	YCbCrPositioning: 1
	ExifOffset: 210
	GPSInfo: 770

Exif: 
	ExposureTime: 1/295 (0.003)
	FNumber: 179/100 (1.79)
	PhotographicSensitivity: 112
	Unknown Tag (0x8895): 0
	ExifVersion: 48, 50, 50, 48
	DateTimeOriginal: '2019:12:16 13:49:17'
	DateTimeDigitized: '2019:12:16 13:49:17'
	ComponentsConfiguration: 1, 2, 3, 0
	ShutterSpeedValue: 8202/1000 (8.202)
	ApertureValue: 167/100 (1.67)
	ExposureCompensation: 0
	MaxApertureValue: 167/100 (1.67)
	MeteringMode: 2
	LightSource: 0
	Flash: 16
	FocalLength: 4740/1000 (4.74)
	SubSecTime: '646073'
	SubSecTimeOriginal: '646073'
	SubSecTimeDigitized: '646073'
	Unknown Tag (0x9999): '{"mirror":false,"sensor_type":"rear","Hdr":"off","OpMode":36869}'
	FlashpixVersion: 48, 49, 48, 48
	ColorSpace: 1
	ExifImageWidth: 4000
	ExifImageLength: 3000
	InteropOffset: 738
	SensingMethod: 1
	WhiteBalance: 0
	FocalLengthIn35mmFormat: 25

Interoperability: 
	InteroperabilityIndex: 'R98'
	InteroperabilityVersion: 48, 49, 48, 48

Gps: 
	GPSLatitudeRef: 'N'
	GPSLatitude: 40, 21, 401435/10000 (40.144)
	GPSLongitudeRef: 'E'
	GPSLongitude: 116, 1, 14916/10000 (1.492)
	GPSAltitudeRef: 0
	GPSAltitude: 722801/1000 (722.801)
	GPSTimeStamp: 5, 49, 12
	GPSProcessingMethod: 'GPS'
	GPSDateStamp: '2019:12:16'
​

我们可以看到该照片是在 2019-12-16 13:49:17 通过小米 CC 9 拍摄，并且可以看到经纬度海拔等信息。

如果没有去除 Exif 里的敏感信息，那么发布到网络上后，任何人都可以查看 Exif 信息，在某些情况下会造成隐私泄露。

Java 操作 Exif

我们可通过 Apache Commons Imaging 来读取和写入 Exif 元数据。

1. 引入依赖 commons-imaging

   <dependency>
     <groupId>org.apache.commons</groupId>
     <artifactId>commons-imaging</artifactId>
     <version>1.0-alpha1</version>
   </dependency>
   ​

2. 读取 Exif

   final ImageMetadata metadata = Imaging.getMetadata(bytes);
   final JpegImageMetadata jpegMetadata = (JpegImageMetadata) metadata;
   final TiffImageMetadata exif = jpegMetadata.getExif();
   ​

3. 写入 Exif

   final TiffOutputSet outputSet = exif.getOutputSet();
   new ExifRewriter().updateExifMetadataLossless(bytes, os, outputSet);
   ​

Java 删除敏感 Exif

这里提供两种删除 Exif 信息的方法，第一种比较暴力，通过图片字节数组中 Exif 标识开头和结尾来完整擦除整个 Exif 段：

public static byte[] removeExif(final byte[] bytes) {
    try {
        final byte b0 = bytes[0];
        final byte b1 = bytes[1];
        if (-1 != b0 || -40 != b1) { // FF D8
            return bytes; // not jpeg
        }

        if (-1 != bytes[2] || -31 != bytes[3]) { // exif seg: FF E1
            return bytes; // no exif
        }

        String len0 = Integer.toHexString(bytes[4]);
        if (2 > len0.length()) {
            len0 = "0" + len0;
        } else {
            len0 = len0.substring(len0.length() - 2);
        }
        String len1 = Integer.toHexString(bytes[5]);
        if (2 > len1.length()) {
            len1 = "0" + len1;
        } else {
            len1 = len1.substring(len1.length() - 2);
        }
        final String lenStr = len0 + "" + len1;
        final int len = Integer.parseInt(lenStr, 16);
        final byte[] ret = new byte[bytes.length - len - 4 - 2];
        ret[0] = -1;
        ret[1] = -40;
        System.arraycopy(bytes, 4 + len, ret, 2, ret.length - 2);
        return ret;
    } catch (final Exception e) {
        LOGGER.log(Level.ERROR, "Removes Exif failed", e);
        return bytes;
    }
}
​

该方法的优点是性能很好，但缺点就是会擦除 Exif 中有用的信息（比如方向 Orientation，如果没有该字段，那么浏览器就无法自动旋转图片为正常方向了）。

另一个方法是通过上面介绍的 commons-imaging 来删除，这样可以保留想要的字段：

public static byte[] removeExif(final byte[] bytes) throws Exception {
    try (final ByteArrayOutputStream os = new ByteArrayOutputStream()) {
        TiffOutputSet outputSet = null;
        final ImageMetadata metadata = Imaging.getMetadata(bytes);
        if (!(metadata instanceof JpegImageMetadata)) {
            return bytes;
        }

        final JpegImageMetadata jpegMetadata = (JpegImageMetadata) metadata;
        final TiffImageMetadata exif = jpegMetadata.getExif();
        if (null != exif) {
            outputSet = exif.getOutputSet();
        }

        if (null == outputSet) {
            return bytes;
        }

        final List<TiffOutputDirectory> directories = outputSet.getDirectories();
        for (final TiffOutputDirectory directory : directories) {
            final List<TiffOutputField> fields = directory.getFields();
            for (final TiffOutputField field : fields) {
                if (!StringUtils.equalsIgnoreCase("Orientation", field.tagInfo.name)) {
                    outputSet.removeField(field.tagInfo);
                }
            }
        }

        new ExifRewriter().updateExifMetadataLossless(bytes, os, outputSet);
        return os.toByteArray();
    }
}
​