如何将多个PNG组合成一个大的PNG文件?
Posted
技术标签:
【中文标题】如何将多个PNG组合成一个大的PNG文件?【英文标题】:How to combine multiple PNGs into one big PNG file? 【发布时间】:2011-04-24 18:00:15 【问题描述】:我有大约。 6000 个 PNG 文件(256*256 像素),并希望以编程方式将它们组合成一个大的 PNG。
最好/最快的方法是什么?
(目的是在纸上打印,所以不能使用一些网络技术,只有一个单一的图片文件可以消除许多使用错误。)
我尝试了 fahd 的建议,但是当我尝试创建一个宽 24576 像素、高 15360 像素的 BufferedImage
时,我得到了一个 NullPointerException
。有什么想法吗?
【问题讨论】:
你想要一个大的PNG?那将是1536000x1536000像素?我必须说一个基本的图像库在这里会是一个更好的选择。 即使有办法绕过重新编码,重新编码的非常大的图像可能会压缩得更好,尤其是在图像相似的情况下。 @kyndigs:更像 15360 x 25600(对于 60 x 100 排列) 无论如何,一个基本的图像库会是一个更好的选择:p ...和一台 300dpi 的 DinA0 打印机,这正是我想要的。 【参考方案1】:创建一个您将写入的大图像。根据您想要的行数和列数计算其尺寸。
BufferedImage result = new BufferedImage(
width, height, //work these out
BufferedImage.TYPE_INT_RGB);
Graphics g = result.getGraphics();
现在循环浏览您的图像并绘制它们:
for(String image : images)
BufferedImage bi = ImageIO.read(new File(image));
g.drawImage(bi, x, y, null);
x += 256;
if(x > result.getWidth())
x = 0;
y += bi.getHeight();
最后写入文件:
ImageIO.write(result,"png",new File("result.png"));
【讨论】:
天哪,这比尝试使用 JAI 和马赛克描述符要容易得多。谢谢!【参考方案2】:前段时间我有一些类似的需求(巨大的图像 - 而且,我的情况是 16 位深度 - 将它们完全存储在内存中不是一种选择)。我结束了对 PNG 库的编码,以按顺序进行读/写。如果有人觉得它有用,那就是here。
更新:这是一个示例代码:
/**
* Takes several tiles and join them in a single image
*
* @param tiles Filenames of PNG files to tile
* @param dest Destination PNG filename
* @param nTilesX How many tiles per row?
*/
public class SampleTileImage
public static void doTiling(String tiles[], String dest, int nTilesX)
int ntiles = tiles.length;
int nTilesY = (ntiles + nTilesX - 1) / nTilesX; // integer ceil
ImageInfo imi1, imi2; // 1:small tile 2:big image
PngReader pngr = new PngReader(new File(tiles[0]));
imi1 = pngr.imgInfo;
PngReader[] readers = new PngReader[nTilesX];
imi2 = new ImageInfo(imi1.cols * nTilesX, imi1.rows * nTilesY, imi1.bitDepth, imi1.alpha, imi1.greyscale,
imi1.indexed);
PngWriter pngw = new PngWriter(new File(dest), imi2, true);
// copy palette and transparency if necessary (more chunks?)
pngw.copyChunksFrom(pngr.getChunksList(), ChunkCopyBehaviour.COPY_PALETTE
| ChunkCopyBehaviour.COPY_TRANSPARENCY);
pngr.readSkippingAllRows(); // reads only metadata
pngr.end(); // close, we'll reopen it again soon
ImageLineInt line2 = new ImageLineInt(imi2);
int row2 = 0;
for (int ty = 0; ty < nTilesY; ty++)
int nTilesXcur = ty < nTilesY - 1 ? nTilesX : ntiles - (nTilesY - 1) * nTilesX;
Arrays.fill(line2.getScanline(), 0);
for (int tx = 0; tx < nTilesXcur; tx++) // open several readers
readers[tx] = new PngReader(new File(tiles[tx + ty * nTilesX]));
readers[tx].setChunkLoadBehaviour(ChunkLoadBehaviour.LOAD_CHUNK_NEVER);
if (!readers[tx].imgInfo.equals(imi1))
throw new RuntimeException("different tile ? " + readers[tx].imgInfo);
for (int row1 = 0; row1 < imi1.rows; row1++, row2++)
for (int tx = 0; tx < nTilesXcur; tx++)
ImageLineInt line1 = (ImageLineInt) readers[tx].readRow(row1); // read line
System.arraycopy(line1.getScanline(), 0, line2.getScanline(), line1.getScanline().length * tx,
line1.getScanline().length);
pngw.writeRow(line2, row2); // write to full image
for (int tx = 0; tx < nTilesXcur; tx++)
readers[tx].end(); // close readers
pngw.end(); // close writer
public static void main(String[] args)
doTiling(new String[] "t1.png", "t2.png", "t3.png", "t4.png", "t5.png", "t6.png" , "tiled.png", 2);
System.out.println("done");
【讨论】:
这很有趣,但是这里的代码与当前版本的库相比已经过时了。你有没有类似的例子是当前的?非常感谢。 @MattFriedman 你试过“片段”中发布的那个吗? code.google.com/p/pngj/wiki/Snippets 您提供的链接上发布的代码已编译。非常感谢。也许你想继续用新的东西替换上面的代码。只是一个想法。再次感谢。【参考方案3】:我看不出“没有处理和重新编码”怎么可能。如果你坚持使用 Java,那么我只建议你使用JAI(项目页面here)。这样你就可以create one big BufferedImage、load smaller images 和draw them on the bigger one。
或者直接使用ImageMagickmontage
:
montage *.png output.png
有关montage
的更多信息,请参阅usage。
【讨论】:
【参考方案4】:PNG 格式不支持平铺,因此您无法逃避至少解压缩和重新压缩数据流。如果所有图像的调色板都相同(或全部不存在),那么这是您真正需要做的唯一事情。 (我还假设图像不是隔行扫描的。)
您可以以流式方式执行此操作,一次只打开一个“行”PNG,从其数据流中读取适当大小的块并将它们写入输出流。这样您就不需要将整个图像保存在内存中。最有效的方法是自己在 libpng 之上进行编程。由于像素预测,您可能需要在内存中保留略多于一条的像素扫描线。
但仅使用 ImageMagick、netpbm 或类似工具的命令行实用程序将为您节省大量开发时间,而收益可能微乎其微。
【讨论】:
【参考方案5】:正如其他人所指出的,使用 Java 不一定是最好的选择。
如果您要使用 Java,最好的选择——假设你的内存足够短,以至于你不能多次将整个数据集读入内存然后再次写出——是实现RenderedImage
有一个类,可以根据需要从磁盘读取您的 PNG。如果您只是创建自己的新 BufferedImage 然后尝试将其写出,PNG 编写器将创建数据的额外副本。如果您创建自己的 RenderedImage,可以将其传递给ImageIO.write(myImageSet,"png",myFileName)
。您可以从您的第一个 PNG 复制 SampleModel
和 ColorModel
信息——希望它们都是一样的。
如果您假设整个图像是多个图块(每个源图像一个图块),那么ImageIO.write
将创建一个WritableRaster
,它是整个图像数据集的大小,并将调用您的@987654327 实现@ 用数据填充它。如果你有足够的内存,这是一个简单的方法(因为你得到了一个巨大的目标数据集,并且可以将所有图像数据转储到其中 - 使用 setRect(dx,dy,Raster)
方法 - 然后不必再次担心它)。我还没有测试过这是否可以节省内存,但在我看来应该可以。
或者,如果您假装整个图像是单个图块,ImageIO.write
将使用getTile(0,0)
询问与整个图像对应的光栅。因此,您必须创建自己的 Raster,这反过来又使您创建自己的 DataBuffer。当我尝试这种方法时,成功写入 15360x25600 RGB PNG 的最小内存使用量是-Xmx1700M
(顺便说一下,在 Scala 中),这只是写入图像的每个像素仅超过 4 个字节,因此在一张完整图像之上几乎没有开销在内存中。
PNG 数据格式本身并不需要将整个图像存储在内存中——它可以在块中正常工作——但遗憾的是,PNG 编写器的默认实现假定它将整个像素数组都存储在内存中。
【讨论】:
【参考方案6】:您最好使用另一种(无损)图像格式来实现。 PPM 非常易于使用(并以编程方式放入磁贴;它只是磁盘上的一个大数组,因此您最多只需要存储一行磁贴),但它非常浪费空间(每个 12 字节像素!)。
然后使用标准转换器(例如ppm2png
),将中间格式转换为巨型 PNG。
【讨论】:
【参考方案7】:用于将图块拼接成一张大图的简单 Python 脚本:
import Image
TILESIZE = 256
ZOOM = 15
def merge_images( xmin, xmax, ymin, ymax, output) :
out = Image.new( 'RGB', ((xmax-xmin+1) * TILESIZE, (ymax-ymin+1) * TILESIZE) )
imx = 0;
for x in range(xmin, xmax+1) :
imy = 0
for y in range(ymin, ymax+1) :
tile = Image.open( "%s_%s_%s.png" % (ZOOM, x, y) )
out.paste( tile, (imx, imy) )
imy += TILESIZE
imx += TILESIZE
out.save( output )
运行:
merge_images(18188, 18207, 11097, 11111, "output.png")
适用于名为 %ZOOM_%XCORD_%YCORD.png 的文件,例如 15_18188_11097.png
【讨论】:
【参考方案8】:组合图片
private static void combineALLImages(String screenNames, int screens) throws IOException, InterruptedException
System.out.println("screenNames --> D:\\screenshots\\screen screens --> 0,1,2 to 10/..");
int rows = screens + 1;
int cols = 1;
int chunks = rows * cols ;
File[] imgFiles = new File[chunks];
String files = "";
for (int i = 0; i < chunks; i++)
files = screenNames + i + ".jpg";
imgFiles[i] = new File(files);
System.out.println(screenNames + i + ".jpg"+"\t Screens : "+screens);
BufferedImage sample = ImageIO.read(imgFiles[0]);
//Initializing the final image
BufferedImage finalImg = new BufferedImage(sample.getWidth() * cols, sample.getHeight() * rows, sample.getType());
int index = 0;
for (int i = 0; i < rows; i++)
for (int j = 0; j < cols; j++)
BufferedImage temp = ImageIO.read(imgFiles[index]);
finalImg.createGraphics().drawImage(temp, sample.getWidth() * j, sample.getHeight() * i, null);
System.out.println(screenNames + index + ".jpg");
index++;
File final_Image = new File("D:\\Screenshots\\FinalImage.jpg");
ImageIO.write(finalImg, "jpeg", final_Image);
【讨论】:
【参考方案9】:我一直回到这个问题,因为我有类似的问题,并在另一个线程中找到了可接受的解决方案,我将在此处链接以供将来参考。
它并不能完全解决 OP 问题,但它确实允许将水平切片(“平铺线”)拼接在一起,而无需使用 AWT API 同时将所有内容加载到内存中。
Merge small images into one without allocating full image in memory
链接的存储库不再可用,但有 mirrors 可用。
/*******************************************************************************
* Copyright (c) MOBAC developers
*
* This program is free software: you can redistribute it and/or modify
* it under the terms of the GNU General Public License as published by
* the Free Software Foundation, either version 2 of the License, or
* (at your option) any later version.
*
* This program is distributed in the hope that it will be useful,
* but WITHOUT ANY WARRANTY; without even the implied warranty of
* MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the
* GNU General Public License for more details.
*
* You should have received a copy of the GNU General Public License
* along with this program. If not, see <http://www.gnu.org/licenses/>.
******************************************************************************/
package mobac.utilities.imageio;
/*
* PNGWriter.java
*
* Copyright (c) 2007 Matthias Mann - www.matthiasmann.de
*
* Permission is hereby granted, free of charge, to any person obtaining a copy
* of this software and associated documentation files (the "Software"), to deal
* in the Software without restriction, including without limitation the rights
* to use, copy, modify, merge, publish, distribute, sublicense, and/or sell
* copies of the Software, and to permit persons to whom the Software is
* furnished to do so, subject to the following conditions:
*
* The above copyright notice and this permission notice shall be included in
* all copies or substantial portions of the Software.
*
* THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
* IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
* FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL THE
* AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER
* LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM,
* OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN
* THE SOFTWARE.
*
*/
import static mobac.utilities.imageio.PngConstants.*;
import java.awt.Rectangle;
import java.awt.image.BufferedImage;
import java.awt.image.ColorModel;
import java.awt.image.DataBuffer;
import java.awt.image.DataBufferByte;
import java.awt.image.DataBufferInt;
import java.awt.image.DirectColorModel;
import java.io.BufferedOutputStream;
import java.io.DataOutputStream;
import java.io.IOException;
import java.io.OutputStream;
import java.util.zip.CRC32;
import java.util.zip.Deflater;
import java.util.zip.DeflaterOutputStream;
import javax.activation.UnsupportedDataTypeException;
/**
* A PNG writer that is able to write extra large PNG images using incremental
* writing.
* <p>
* The image is processed incremental in "tile lines" - e.g. an PNG image of
* 30000 x 20000 pixels (width x height) can be written by 200 "tile lines" of
* size 30000 x 100 pixels. Each tile line can be written via the method
* @link #writeTileLine(BufferedImage). After writing the last line you have
* to call @link #finish() which will write the final PNG structure
* information into the @link OutputStream.
* </p>
* <p>
* Please note that this writer creates 24bit/truecolor PNGs. Transparency and
* alpha masks are not supported.
* </p>
* Bases on the PNGWriter written by Matthias Mann - www.matthiasmann.de
*
* @author r_x
*/
public class PngXxlWriter
private static final int BUFFER_SIZE = 128 * 1024;
private int width;
private int height;
private DataOutputStream dos;
ImageDataChunkWriter imageDataChunkWriter;
/**
* Creates an PNG writer instance for an image with the specified width and
* height.
*
* @param width
* width of the PNG image to be written
* @param height
* height of the PNG image to be written
* @param os
* destination to write the PNG image data to
* @throws IOException
*/
public PngXxlWriter(int width, int height, OutputStream os) throws IOException
this.width = width;
this.height = height;
this.dos = new DataOutputStream(os);
dos.write(SIGNATURE);
PngChunk cIHDR = new PngChunk(IHDR);
cIHDR.writeInt(this.width);
cIHDR.writeInt(this.height);
cIHDR.writeByte(8); // 8 bit per component
cIHDR.writeByte(COLOR_TRUECOLOR);
cIHDR.writeByte(COMPRESSION_DEFLATE);
cIHDR.writeByte(FILTER_SET_1);
cIHDR.writeByte(INTERLACE_NONE);
cIHDR.writeTo(dos);
imageDataChunkWriter = new ImageDataChunkWriter(dos);
/**
*
* @param tileLineImage
* @throws IOException
*/
public void writeTileLine(BufferedImage tileLineImage) throws IOException
int tileLineHeight = tileLineImage.getHeight();
int tileLineWidth = tileLineImage.getWidth();
if (width != tileLineWidth)
throw new RuntimeException("Invalid width");
ColorModel cm = tileLineImage.getColorModel();
if (!(cm instanceof DirectColorModel))
throw new UnsupportedDataTypeException(
"Image uses wrong color model. Only DirectColorModel is supported!");
// We process the image line by line, from head to bottom
Rectangle rect = new Rectangle(0, 0, tileLineWidth, 1);
DataOutputStream imageDataStream = imageDataChunkWriter.getStream();
byte[] curLine = new byte[width * 3];
for (int line = 0; line < tileLineHeight; line++)
rect.y = line;
DataBuffer db = tileLineImage.getData(rect).getDataBuffer();
if (db.getNumBanks() > 1)
throw new UnsupportedDataTypeException("Image data has more than one data bank");
if (db instanceof DataBufferByte)
curLine = ((DataBufferByte) db).getData();
else if (db instanceof DataBufferInt)
int[] intLine = ((DataBufferInt) db).getData();
int c = 0;
for (int i = 0; i < intLine.length; i++)
int pixel = intLine[i];
curLine[c++] = (byte) (pixel >> 16 & 0xFF);
curLine[c++] = (byte) (pixel >> 8 & 0xFF);
curLine[c++] = (byte) (pixel & 0xFF);
else
throw new UnsupportedDataTypeException(db.getClass().getName());
imageDataStream.write(FILTER_TYPE_NONE);
imageDataStream.write(curLine);
public void finish() throws IOException
imageDataChunkWriter.finish();
PngChunk cIEND = new PngChunk(IEND);
cIEND.writeTo(dos);
cIEND.close();
dos.flush();
static class ImageDataChunkWriter extends OutputStream
DeflaterOutputStream dfos;
DataOutputStream stream;
DataOutputStream out;
CRC32 crc = new CRC32();
public ImageDataChunkWriter(DataOutputStream out) throws IOException
this.out = out;
dfos = new DeflaterOutputStream(new BufferedOutputStream(this, BUFFER_SIZE),
new Deflater(Deflater.BEST_COMPRESSION));
stream = new DataOutputStream(dfos);
public DataOutputStream getStream()
return stream;
public void finish() throws IOException
stream.flush();
stream.close();
dfos.finish();
dfos = null;
@Override
public void write(byte[] b, int off, int len) throws IOException
crc.reset();
out.writeInt(len);
out.writeInt(IDAT);
out.write(b, off, len);
crc.update("IDAT".getBytes());
crc.update(b, off, len);
out.writeInt((int) crc.getValue());
@Override
public void write(byte[] b) throws IOException
write(b, 0, b.length);
@Override
public void write(int b) throws IOException
throw new IOException("Simgle byte writing not supported");
【讨论】:
【参考方案10】:像这样使用 imagemagick 的蒙太奇:
montage *.png montage.png
您可以找到更多关于参数here的信息
祝你好运
【讨论】:
以上是关于如何将多个PNG组合成一个大的PNG文件?的主要内容,如果未能解决你的问题,请参考以下文章
AngularJS,如何将多个输入组合成一个 ng 模型?具体来说,将日期、时间和时区输入组合到 datetime 对象中