HybridBinarizer.java revision 7e3fa36d69ffee874dd364b8e3d9aa3cab9a273b
/*
* Copyright 2009 ZXing authors
*
* Licensed under the Apache License, Version 2.0 (the "License");
* you may not use this file except in compliance with the License.
* You may obtain a copy of the License at
*
*
* Unless required by applicable law or agreed to in writing, software
* distributed under the License is distributed on an "AS IS" BASIS,
* WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
* See the License for the specific language governing permissions and
* limitations under the License.
*/
/**
* This class implements a local thresholding algorithm, which while slower than the
* GlobalHistogramBinarizer, is fairly efficient for what it does. It is designed for
* high frequency images of barcodes with black data on white backgrounds. For this application,
* it does a much better job than a global blackpoint with severe shadows and gradients.
* However it tends to produce artifacts on lower frequency images and is therefore not
* a good general purpose binarizer for uses outside ZXing.
*
* This class extends GlobalHistogramBinarizer, using the older histogram approach for 1D readers,
* and the newer local approach for 2D readers. 1D decoding using a per-row histogram is already
* inherently local, and only fails for horizontal gradients. We can revisit that problem later,
* but for now it was not a win to use local blocks for 1D.
*
* This Binarizer is the default for the unit tests and the recommended class for library users.
*
* @author dswitkin@google.com (Daniel Switkin)
*/
public final class HybridBinarizer extends GlobalHistogramBinarizer {
// This class uses 5x5 blocks to compute local luminance, where each block is 8x8 pixels.
// So this is the smallest dimension in each axis we can accept.
private static final int BLOCK_SIZE_POWER = 3;
private static final int MIN_DYNAMIC_RANGE = 24;
super(source);
}
/**
* Calculates the final BitMatrix once for all requests. This could be called once from the
* constructor instead, but there are some advantages to doing it lazily, such as making
* profiling easier, and not doing heavy lifting when callers don't expect it.
*/
return matrix;
}
subWidth++;
}
subHeight++;
}
} else {
// If the image is too small, fall back to the global histogram approach.
matrix = super.getBlackMatrix();
}
return matrix;
}
return new HybridBinarizer(source);
}
/**
* For each block in the image, calculate the average black point using a 5x5 grid
* of the blocks around it. Also handles the corner cases (fractional blocks are computed based
*/
private static void calculateThresholdForBlock(byte[] luminances,
int subWidth,
int subHeight,
int width,
int height,
int[][] blackPoints,
for (int y = 0; y < subHeight; y++) {
int yoffset = y << BLOCK_SIZE_POWER;
if (yoffset > maxYOffset) {
}
for (int x = 0; x < subWidth; x++) {
int xoffset = x << BLOCK_SIZE_POWER;
if (xoffset > maxXOffset) {
}
int sum = 0;
for (int z = -2; z <= 2; z++) {
sum += blackRow[left - 2] + blackRow[left - 1] + blackRow[left] + blackRow[left + 1] + blackRow[left + 2];
}
}
}
}
}
/**
* Applies a single threshold to a block of pixels.
*/
private static void thresholdBlock(byte[] luminances,
int xoffset,
int yoffset,
int threshold,
int stride,
for (int x = 0; x < BLOCK_SIZE; x++) {
// Comparison needs to be <= so that black == 0 pixels are black even if the threshold is 0.
}
}
}
}
/**
* Calculates a single black point for each block of pixels and saves it away.
* See the following thread for a discussion of this algorithm:
*/
private static int[][] calculateBlackPoints(byte[] luminances,
int subWidth,
int subHeight,
int width,
int height) {
for (int y = 0; y < subHeight; y++) {
int yoffset = y << BLOCK_SIZE_POWER;
if (yoffset > maxYOffset) {
}
for (int x = 0; x < subWidth; x++) {
int xoffset = x << BLOCK_SIZE_POWER;
if (xoffset > maxXOffset) {
}
int sum = 0;
int min = 0xFF;
int max = 0;
// still looking for good contrast
}
}
}
// finish the rest of the rows quickly
}
}
}
}
// The default estimate is the average of the values in the block.
// If variation within the block is low, assume this is a block with only light or only
// dark pixels. In that case we do not want to use the average, as it would divide this
// low contrast area into black and white pixels, essentially creating data out of noise.
//
// The default assumption is that the block is light/background. Since no estimate for
// the level of dark pixels exists locally, use half the min for the block.
if (y > 0 && x > 0) {
// Correct the "white background" assumption for blocks that have neighbors by comparing
// the pixels in this block to the previously calculated black points. This is based on
// the fact that dark barcode symbology is always surrounded by some amount of light
// background for which reasonable black point estimates were made. The bp estimated at
// the boundaries is used for the interior.
// The (min < bp) is arbitrary but works better than other heuristics that were tried.
if (min < averageNeighborBlackPoint) {
}
}
}
blackPoints[y][x] = average;
}
}
return blackPoints;
}
}