Skip to content

Commit

Permalink
rework LeptonicaFrameConverter#convert(Frame) to handle non-aligned s…
Browse files Browse the repository at this point in the history 
…tride sizes, see issue #1115)
  • Loading branch information
@msmuenchen
msmuenchen committed Jan 27, 2024
1 parent dac0b8a commit 2bf5cc9
Showing 1 changed file with 96 additions and 32 deletions.
128 changes: 96 additions & 32 deletions src/main/java/org/bytedeco/javacv/LeptonicaFrameConverter.java
View file
Original file line number Diff line number Diff line change
Expand Up @@ -22,14 +22,16 @@

package org.bytedeco.javacv;

import java.nio.Buffer;
import java.nio.ByteBuffer;
import java.nio.ByteOrder;
import org.bytedeco.javacpp.BytePointer;
import org.bytedeco.javacpp.Loader;
import org.bytedeco.javacpp.Pointer;
import org.bytedeco.leptonica.PIX;

import java.nio.Buffer;
import java.nio.ByteBuffer;
import java.nio.ByteOrder;
import java.nio.IntBuffer;

import org.bytedeco.leptonica.*;
import static org.bytedeco.leptonica.global.leptonica.*;

/**
Expand All @@ -40,7 +42,9 @@
* @author Samuel Audet
*/
public class LeptonicaFrameConverter extends FrameConverter<PIX> {
static { Loader.load(org.bytedeco.leptonica.global.leptonica.class); }
static {
Loader.load(org.bytedeco.leptonica.global.leptonica.class);
}

PIX pix;
BytePointer frameData, pixData;
Expand All @@ -51,40 +55,99 @@ static boolean isEqual(Frame frame, PIX pix) {
&& frame.imageWidth == pix.w() && frame.imageHeight == pix.h()
&& frame.imageChannels == pix.d() / 8 && frame.imageDepth == Frame.DEPTH_UBYTE
&& (ByteOrder.nativeOrder().equals(ByteOrder.LITTLE_ENDIAN)
|| new Pointer(frame.image[0]).address() == pix.data().address())
|| new Pointer(frame.image[0]).address() == pix.data().address())
&& frame.imageStride * Math.abs(frame.imageDepth) / 8 == pix.wpl() * 4;
}

public PIX convert(Frame frame) {
if (frame == null || frame.image == null) {
return null;
} else if (frame.opaque instanceof PIX) {
return (PIX)frame.opaque;
return (PIX) frame.opaque;
} else if (!isEqual(frame, pix)) {
Pointer data;
if (ByteOrder.nativeOrder().equals(ByteOrder.LITTLE_ENDIAN)) {
if (pixData == null || pixData.capacity() < frame.imageHeight * frame.imageStride) {
if (pixData != null) {
pixData.releaseReference();
//I simply lack a machine to test this.
if (ByteOrder.nativeOrder().equals(ByteOrder.BIG_ENDIAN)) {
System.err.println("This converter does not support running on big-endian machines");
return null;
}
//PIX data should be packed as tightly as possible, see https://github.com/DanBloomberg/leptonica/blob/0d4477653691a8cb4f63fa751d43574c757ccc9f/src/pix.h#L133
//For anything not greyscale or RGB @ 8 bit per pixel, this involves more bit-shift logic than I'm willing to write (and I lack test cases)
if (frame.imageChannels != 3 && frame.imageChannels != 1) {
System.out.println(String.format("Image has %d channels, converter only supports 3 (RGB) or 1 (grayscale) for input", frame.imageChannels));
return null;
}
if (frame.imageDepth != 8 || !(frame.image[0] instanceof ByteBuffer)) {
System.out.println(String.format("Image has bit depth %d, converter only supports 8 (1 byte/px) for input", frame.imageDepth));
return null;
}
// Leptonica frame scan lines must be padded to 32 bit / 4 bytes of stride (line) length, otherwise one gets nasty scan effects
// See http://www.leptonica.org/library-notes.html#PIX
int srcChannelDepthBytes = frame.imageDepth / 8;
int srcBytesPerPixel = srcChannelDepthBytes * frame.imageChannels;

// Leptonica counts RGB images as 24 bits per pixel, while the data actually is 32 bit per pixel
int destBytesPerPixel = srcBytesPerPixel;
if (frame.imageChannels == 3) {
// RGB pixels are stored as RGBA, so they take up 4 bytes!
// See https://github.com/DanBloomberg/leptonica/blob/master/src/pix.h#L157
destBytesPerPixel = 4;
}
int currentStrideLength = frame.imageWidth * destBytesPerPixel;
int targetStridePad = 4 - (currentStrideLength % 4);
if (targetStridePad == 4)
targetStridePad = 0;
int targetStrideLength = (currentStrideLength) + targetStridePad;
ByteBuffer src = ((ByteBuffer) frame.image[0]).order(ByteOrder.LITTLE_ENDIAN);
int newSize = targetStrideLength * frame.imageHeight;
ByteBuffer dst = ByteBuffer.allocate(newSize).order(ByteOrder.LITTLE_ENDIAN);
/*
System.out.println(String.format(
"src: %d bytes total, %d channels @ %d bytes per pixel, stride length %d",
frame.image[0].capacity(),
frame.imageChannels,
srcBytesPerPixel,
currentStrideLength
));
System.out.println(String.format(
"dst: %d bytes total, stride length %d, stride pad %d",
newSize,
targetStrideLength,
targetStridePad
));
*/
//The source bytes will be RGB, which means it will have to be copied byte-by-byte to match Leptonica RGBA
//todo: use qword copy ops?
byte[] rowData = new byte[targetStrideLength];
for (int row = 0; row < frame.imageHeight; row++) {
for (int col = 0; col < frame.imageWidth; col++) {
int srcIndex = (frame.imageStride * row) + (col * frame.imageChannels);
if (frame.imageChannels == 1) {
byte v = src.get(srcIndex);
rowData[col] = v;
//System.out.println(String.format("row %03d col %03d idx src %03d val %02x", row, col, srcIndex,v));
} else if (frame.imageChannels == 3) {
int dstIndex = col * destBytesPerPixel;
byte[] pixelData = new byte[3];
src.position(srcIndex);
src.get(pixelData, 0, pixelData.length);
// Convert BGR (OpenCV's standard ordering) to RGB (Leptonica)
// See https://learnopencv.com/why-does-opencv-use-bgr-color-format/ and https://github.com/DanBloomberg/leptonica/blob/master/src/pix.h#L157
rowData[dstIndex] = pixelData[2]; //dst: r
rowData[dstIndex + 1] = pixelData[1]; //dst: g
rowData[dstIndex + 2] = pixelData[0]; // dst: b
rowData[dstIndex + 3] = 0;
//System.out.println(String.format("row %03d col %03d idx src %03d dst %03d val r %02x g %02x b %02x", row, col, srcIndex,dstIndex, pixelData[2], pixelData[1], pixelData[1]));
}
pixData = new BytePointer(frame.imageHeight * frame.imageStride).retainReference();
}
data = pixData;
pixBuffer = data.asByteBuffer().order(ByteOrder.BIG_ENDIAN);
} else {
data = new Pointer(frame.image[0].position(0));
}
if (pix != null) {
pix.releaseReference();
//And since pixel data in source is little-endian, but Leptonica is big-endian on 32-bit level, now invert accordingly...
ByteBuffer rowBuffer = ByteBuffer.wrap(rowData);
IntBuffer inverted = rowBuffer.order(ByteOrder.BIG_ENDIAN).asIntBuffer();
//System.out.println(Arrays.toString(rowBuffer.array()));
dst.position(row * targetStrideLength).asIntBuffer().put(inverted);
}
pix = PIX.create(frame.imageWidth, frame.imageHeight, frame.imageChannels * 8, data)
.wpl(frame.imageStride / 4 * Math.abs(frame.imageDepth) / 8).retainReference();
pix = PIX.create(frame.imageWidth, frame.imageHeight, destBytesPerPixel * 8, new BytePointer(dst.position(0)));
}

if (ByteOrder.nativeOrder().equals(ByteOrder.LITTLE_ENDIAN)) {
((ByteBuffer)pixBuffer.position(0)).asIntBuffer()
.put(((ByteBuffer)frame.image[0].position(0)).order(ByteOrder.LITTLE_ENDIAN).asIntBuffer());
}
return pix;
}

Expand All @@ -99,10 +162,10 @@ public Frame convert(PIX pix) {
} else if (pix.d() < 8) {
switch (pix.d()) {
case 1:
tempPix = pix = pixConvert1To8(null, pix, (byte)0, (byte)255);
tempPix = pix = pixConvert1To8(null, pix, (byte) 0, (byte) 255);
break;
case 2:
tempPix = pix = pixConvert2To8(pix, (byte)0, (byte)85, (byte)170, (byte)255, 0);
tempPix = pix = pixConvert2To8(pix, (byte) 0, (byte) 85, (byte) 170, (byte) 255, 0);
break;
case 4:
tempPix = pix = pixConvert4To8(pix, 0);
Expand All @@ -128,7 +191,7 @@ public Frame convert(PIX pix) {
}
frameBuffer = frameData.asByteBuffer().order(ByteOrder.LITTLE_ENDIAN);
frame.opaque = frameData;
frame.image = new Buffer[] { frameBuffer };
frame.image = new Buffer[]{frameBuffer};
} else {
if (tempPix != null) {
if (this.pix != null) {
Expand All @@ -137,12 +200,12 @@ public Frame convert(PIX pix) {
this.pix = pix = pix.clone();
}
frame.opaque = pix;
frame.image = new Buffer[] { pix.createBuffer() };
frame.image = new Buffer[]{pix.createBuffer()};
}
}

if (ByteOrder.nativeOrder().equals(ByteOrder.LITTLE_ENDIAN)) {
((ByteBuffer)frameBuffer.position(0)).asIntBuffer()
((ByteBuffer) frameBuffer.position(0)).asIntBuffer()
.put(pix.createBuffer().order(ByteOrder.BIG_ENDIAN).asIntBuffer());
}

Expand All @@ -152,7 +215,8 @@ public Frame convert(PIX pix) {
return frame;
}

@Override public void close() {
@Override
public void close() {
super.close();
if (pix != null) {
pix.releaseReference();
Expand Down

0 comments on commit 2bf5cc9

Please sign in to comment.