From: Nigel Tao Date: Tue, 19 Apr 2011 01:00:47 +0000 (+1000) Subject: image/jpeg: add an encoder. X-Git-Tag: weekly.2011-04-27~109 X-Git-Url: http://www.git.cypherpunks.su/?a=commitdiff_plain;h=5500f027f72cc7467e7f30447f83480448674f67;p=gostls13.git image/jpeg: add an encoder. It is based on changeset 4186064 by Raph Levien . R=r, nigeltao_gnome CC=golang-dev https://golang.org/cl/4435051 --- diff --git a/src/pkg/Makefile b/src/pkg/Makefile index de96229723..d3ec7dd290 100644 --- a/src/pkg/Makefile +++ b/src/pkg/Makefile @@ -183,7 +183,6 @@ NOTEST+=\ hash\ http/pprof\ http/httptest\ - image/jpeg\ net/dict\ rand\ runtime/cgo\ diff --git a/src/pkg/image/jpeg/Makefile b/src/pkg/image/jpeg/Makefile index 5c5f97e718..d9d830f2ff 100644 --- a/src/pkg/image/jpeg/Makefile +++ b/src/pkg/image/jpeg/Makefile @@ -6,8 +6,10 @@ include ../../../Make.inc TARG=image/jpeg GOFILES=\ + fdct.go\ huffman.go\ idct.go\ reader.go\ + writer.go\ include ../../../Make.pkg diff --git a/src/pkg/image/jpeg/fdct.go b/src/pkg/image/jpeg/fdct.go new file mode 100644 index 0000000000..3f8be4e326 --- /dev/null +++ b/src/pkg/image/jpeg/fdct.go @@ -0,0 +1,190 @@ +// Copyright 2011 The Go Authors. All rights reserved. +// Use of this source code is governed by a BSD-style +// license that can be found in the LICENSE file. + +package jpeg + +// This file implements a Forward Discrete Cosine Transformation. + +/* +It is based on the code in jfdctint.c from the Independent JPEG Group, +found at http://www.ijg.org/files/jpegsrc.v8c.tar.gz. + +The "LEGAL ISSUES" section of the README in that archive says: + +In plain English: + +1. We don't promise that this software works. (But if you find any bugs, + please let us know!) +2. You can use this software for whatever you want. You don't have to pay us. +3. You may not pretend that you wrote this software. If you use it in a + program, you must acknowledge somewhere in your documentation that + you've used the IJG code. + +In legalese: + +The authors make NO WARRANTY or representation, either express or implied, +with respect to this software, its quality, accuracy, merchantability, or +fitness for a particular purpose. This software is provided "AS IS", and you, +its user, assume the entire risk as to its quality and accuracy. + +This software is copyright (C) 1991-2011, Thomas G. Lane, Guido Vollbeding. +All Rights Reserved except as specified below. + +Permission is hereby granted to use, copy, modify, and distribute this +software (or portions thereof) for any purpose, without fee, subject to these +conditions: +(1) If any part of the source code for this software is distributed, then this +README file must be included, with this copyright and no-warranty notice +unaltered; and any additions, deletions, or changes to the original files +must be clearly indicated in accompanying documentation. +(2) If only executable code is distributed, then the accompanying +documentation must state that "this software is based in part on the work of +the Independent JPEG Group". +(3) Permission for use of this software is granted only if the user accepts +full responsibility for any undesirable consequences; the authors accept +NO LIABILITY for damages of any kind. + +These conditions apply to any software derived from or based on the IJG code, +not just to the unmodified library. If you use our work, you ought to +acknowledge us. + +Permission is NOT granted for the use of any IJG author's name or company name +in advertising or publicity relating to this software or products derived from +it. This software may be referred to only as "the Independent JPEG Group's +software". + +We specifically permit and encourage the use of this software as the basis of +commercial products, provided that all warranty or liability claims are +assumed by the product vendor. +*/ + +// Trigonometric constants in 13-bit fixed point format. +const ( + fix_0_298631336 = 2446 + fix_0_390180644 = 3196 + fix_0_541196100 = 4433 + fix_0_765366865 = 6270 + fix_0_899976223 = 7373 + fix_1_175875602 = 9633 + fix_1_501321110 = 12299 + fix_1_847759065 = 15137 + fix_1_961570560 = 16069 + fix_2_053119869 = 16819 + fix_2_562915447 = 20995 + fix_3_072711026 = 25172 +) + +const ( + constBits = 13 + pass1Bits = 2 + centerJSample = 128 +) + +// fdct performs a forward DCT on an 8x8 block of coefficients, including a +// level shift. +func fdct(b *block) { + // Pass 1: process rows. + for y := 0; y < 8; y++ { + x0 := b[y*8+0] + x1 := b[y*8+1] + x2 := b[y*8+2] + x3 := b[y*8+3] + x4 := b[y*8+4] + x5 := b[y*8+5] + x6 := b[y*8+6] + x7 := b[y*8+7] + + tmp0 := x0 + x7 + tmp1 := x1 + x6 + tmp2 := x2 + x5 + tmp3 := x3 + x4 + + tmp10 := tmp0 + tmp3 + tmp12 := tmp0 - tmp3 + tmp11 := tmp1 + tmp2 + tmp13 := tmp1 - tmp2 + + tmp0 = x0 - x7 + tmp1 = x1 - x6 + tmp2 = x2 - x5 + tmp3 = x3 - x4 + + b[y*8+0] = (tmp10 + tmp11 - 8*centerJSample) << pass1Bits + b[y*8+4] = (tmp10 - tmp11) << pass1Bits + z1 := (tmp12 + tmp13) * fix_0_541196100 + z1 += 1 << (constBits - pass1Bits - 1) + b[y*8+2] = (z1 + tmp12*fix_0_765366865) >> (constBits - pass1Bits) + b[y*8+6] = (z1 - tmp13*fix_1_847759065) >> (constBits - pass1Bits) + + tmp10 = tmp0 + tmp3 + tmp11 = tmp1 + tmp2 + tmp12 = tmp0 + tmp2 + tmp13 = tmp1 + tmp3 + z1 = (tmp12 + tmp13) * fix_1_175875602 + z1 += 1 << (constBits - pass1Bits - 1) + tmp0 = tmp0 * fix_1_501321110 + tmp1 = tmp1 * fix_3_072711026 + tmp2 = tmp2 * fix_2_053119869 + tmp3 = tmp3 * fix_0_298631336 + tmp10 = tmp10 * -fix_0_899976223 + tmp11 = tmp11 * -fix_2_562915447 + tmp12 = tmp12 * -fix_0_390180644 + tmp13 = tmp13 * -fix_1_961570560 + + tmp12 += z1 + tmp13 += z1 + b[y*8+1] = (tmp0 + tmp10 + tmp12) >> (constBits - pass1Bits) + b[y*8+3] = (tmp1 + tmp11 + tmp13) >> (constBits - pass1Bits) + b[y*8+5] = (tmp2 + tmp11 + tmp12) >> (constBits - pass1Bits) + b[y*8+7] = (tmp3 + tmp10 + tmp13) >> (constBits - pass1Bits) + } + // Pass 2: process columns. + // We remove pass1Bits scaling, but leave results scaled up by an overall factor of 8. + for x := 0; x < 8; x++ { + tmp0 := b[0*8+x] + b[7*8+x] + tmp1 := b[1*8+x] + b[6*8+x] + tmp2 := b[2*8+x] + b[5*8+x] + tmp3 := b[3*8+x] + b[4*8+x] + + tmp10 := tmp0 + tmp3 + 1<<(pass1Bits-1) + tmp12 := tmp0 - tmp3 + tmp11 := tmp1 + tmp2 + tmp13 := tmp1 - tmp2 + + tmp0 = b[0*8+x] - b[7*8+x] + tmp1 = b[1*8+x] - b[6*8+x] + tmp2 = b[2*8+x] - b[5*8+x] + tmp3 = b[3*8+x] - b[4*8+x] + + b[0*8+x] = (tmp10 + tmp11) >> pass1Bits + b[4*8+x] = (tmp10 - tmp11) >> pass1Bits + + z1 := (tmp12 + tmp13) * fix_0_541196100 + z1 += 1 << (constBits + pass1Bits - 1) + b[2*8+x] = (z1 + tmp12*fix_0_765366865) >> (constBits + pass1Bits) + b[6*8+x] = (z1 - tmp13*fix_1_847759065) >> (constBits + pass1Bits) + + tmp10 = tmp0 + tmp3 + tmp11 = tmp1 + tmp2 + tmp12 = tmp0 + tmp2 + tmp13 = tmp1 + tmp3 + z1 = (tmp12 + tmp13) * fix_1_175875602 + z1 += 1 << (constBits + pass1Bits - 1) + tmp0 = tmp0 * fix_1_501321110 + tmp1 = tmp1 * fix_3_072711026 + tmp2 = tmp2 * fix_2_053119869 + tmp3 = tmp3 * fix_0_298631336 + tmp10 = tmp10 * -fix_0_899976223 + tmp11 = tmp11 * -fix_2_562915447 + tmp12 = tmp12 * -fix_0_390180644 + tmp13 = tmp13 * -fix_1_961570560 + + tmp12 += z1 + tmp13 += z1 + b[1*8+x] = (tmp0 + tmp10 + tmp12) >> (constBits + pass1Bits) + b[3*8+x] = (tmp1 + tmp11 + tmp13) >> (constBits + pass1Bits) + b[5*8+x] = (tmp2 + tmp11 + tmp12) >> (constBits + pass1Bits) + b[7*8+x] = (tmp3 + tmp10 + tmp13) >> (constBits + pass1Bits) + } +} diff --git a/src/pkg/image/jpeg/idct.go b/src/pkg/image/jpeg/idct.go index 5189931105..e5a2f40f5d 100644 --- a/src/pkg/image/jpeg/idct.go +++ b/src/pkg/image/jpeg/idct.go @@ -63,7 +63,7 @@ const ( // // For more on the actual algorithm, see Z. Wang, "Fast algorithms for the discrete W transform and // for the discrete Fourier transform", IEEE Trans. on ASSP, Vol. ASSP- 32, pp. 803-816, Aug. 1984. -func idct(b *[blockSize]int) { +func idct(b *block) { // Horizontal 1-D IDCT. for y := 0; y < 8; y++ { // If all the AC components are zero, then the IDCT is trivial. diff --git a/src/pkg/image/jpeg/reader.go b/src/pkg/image/jpeg/reader.go index fb9cb11bb7..63b594e91c 100644 --- a/src/pkg/image/jpeg/reader.go +++ b/src/pkg/image/jpeg/reader.go @@ -2,11 +2,11 @@ // Use of this source code is governed by a BSD-style // license that can be found in the LICENSE file. -// The jpeg package implements a decoder for JPEG images, as defined in ITU-T T.81. +// Package jpeg implements a JPEG image decoder and encoder. +// +// JPEG is defined in ITU-T T.81: http://www.w3.org/Graphics/JPEG/itu-t81.pdf. package jpeg -// See http://www.w3.org/Graphics/JPEG/itu-t81.pdf - import ( "bufio" "image" @@ -32,6 +32,8 @@ type component struct { tq uint8 // Quantization table destination selector. } +type block [blockSize]int + const ( blockSize = 64 // A DCT block is 8x8. @@ -88,9 +90,9 @@ type decoder struct { ri int // Restart Interval. comps [nComponent]component huff [maxTc + 1][maxTh + 1]huffman - quant [maxTq + 1][blockSize]int + quant [maxTq + 1]block b bits - blocks [nComponent][maxH * maxV][blockSize]int + blocks [nComponent][maxH * maxV]block tmp [1024]byte } @@ -269,7 +271,7 @@ func (d *decoder) processSOS(n int) os.Error { myy := (d.height + 8*int(v0) - 1) / (8 * int(v0)) mcu, expectedRST := 0, uint8(rst0Marker) - var allZeroes [blockSize]int + var allZeroes block var dc [nComponent]int for my := 0; my < myy; my++ { for mx := 0; mx < mxx; mx++ { diff --git a/src/pkg/image/jpeg/writer.go b/src/pkg/image/jpeg/writer.go new file mode 100644 index 0000000000..505cce04fb --- /dev/null +++ b/src/pkg/image/jpeg/writer.go @@ -0,0 +1,523 @@ +// Copyright 2011 The Go Authors. All rights reserved. +// Use of this source code is governed by a BSD-style +// license that can be found in the LICENSE file. + +package jpeg + +import ( + "bufio" + "image" + "image/ycbcr" + "io" + "os" +) + +// min returns the minimum of two integers. +func min(x, y int) int { + if x < y { + return x + } + return y +} + +// div returns a/b rounded to the nearest integer, instead of rounded to zero. +func div(a int, b int) int { + if a >= 0 { + return (a + (b >> 1)) / b + } + return -((-a + (b >> 1)) / b) +} + +// bitCount counts the number of bits needed to hold an integer. +var bitCount = [256]byte{ + 0, 1, 2, 2, 3, 3, 3, 3, 4, 4, 4, 4, 4, 4, 4, 4, + 5, 5, 5, 5, 5, 5, 5, 5, 5, 5, 5, 5, 5, 5, 5, 5, + 6, 6, 6, 6, 6, 6, 6, 6, 6, 6, 6, 6, 6, 6, 6, 6, + 6, 6, 6, 6, 6, 6, 6, 6, 6, 6, 6, 6, 6, 6, 6, 6, + 7, 7, 7, 7, 7, 7, 7, 7, 7, 7, 7, 7, 7, 7, 7, 7, + 7, 7, 7, 7, 7, 7, 7, 7, 7, 7, 7, 7, 7, 7, 7, 7, + 7, 7, 7, 7, 7, 7, 7, 7, 7, 7, 7, 7, 7, 7, 7, 7, + 7, 7, 7, 7, 7, 7, 7, 7, 7, 7, 7, 7, 7, 7, 7, 7, + 8, 8, 8, 8, 8, 8, 8, 8, 8, 8, 8, 8, 8, 8, 8, 8, + 8, 8, 8, 8, 8, 8, 8, 8, 8, 8, 8, 8, 8, 8, 8, 8, + 8, 8, 8, 8, 8, 8, 8, 8, 8, 8, 8, 8, 8, 8, 8, 8, + 8, 8, 8, 8, 8, 8, 8, 8, 8, 8, 8, 8, 8, 8, 8, 8, + 8, 8, 8, 8, 8, 8, 8, 8, 8, 8, 8, 8, 8, 8, 8, 8, + 8, 8, 8, 8, 8, 8, 8, 8, 8, 8, 8, 8, 8, 8, 8, 8, + 8, 8, 8, 8, 8, 8, 8, 8, 8, 8, 8, 8, 8, 8, 8, 8, + 8, 8, 8, 8, 8, 8, 8, 8, 8, 8, 8, 8, 8, 8, 8, 8, +} + +type quantIndex int + +const ( + quantIndexLuminance quantIndex = iota + quantIndexChrominance + nQuantIndex +) + +// unscaledQuant are the unscaled quantization tables. Each encoder copies and +// scales the tables according to its quality parameter. +var unscaledQuant = [nQuantIndex][blockSize]byte{ + // Luminance. + { + 16, 11, 10, 16, 24, 40, 51, 61, + 12, 12, 14, 19, 26, 58, 60, 55, + 14, 13, 16, 24, 40, 57, 69, 56, + 14, 17, 22, 29, 51, 87, 80, 62, + 18, 22, 37, 56, 68, 109, 103, 77, + 24, 35, 55, 64, 81, 104, 113, 92, + 49, 64, 78, 87, 103, 121, 120, 101, + 72, 92, 95, 98, 112, 100, 103, 99, + }, + // Chrominance. + { + 17, 18, 24, 47, 99, 99, 99, 99, + 18, 21, 26, 66, 99, 99, 99, 99, + 24, 26, 56, 99, 99, 99, 99, 99, + 47, 66, 99, 99, 99, 99, 99, 99, + 99, 99, 99, 99, 99, 99, 99, 99, + 99, 99, 99, 99, 99, 99, 99, 99, + 99, 99, 99, 99, 99, 99, 99, 99, + 99, 99, 99, 99, 99, 99, 99, 99, + }, +} + +type huffIndex int + +const ( + huffIndexLuminanceDC huffIndex = iota + huffIndexLuminanceAC + huffIndexChrominanceDC + huffIndexChrominanceAC + nHuffIndex +) + +// huffmanSpec specifies a Huffman encoding. +type huffmanSpec struct { + // count[i] is the number of codes of length i bits. + count [16]byte + // value[i] is the decoded value of the i'th codeword. + value []byte +} + +// theHuffmanSpec is the Huffman encoding specifications. +// This encoder uses the same Huffman encoding for all images. +var theHuffmanSpec = [nHuffIndex]huffmanSpec{ + // Luminance DC. + { + [16]byte{0, 1, 5, 1, 1, 1, 1, 1, 1, 0, 0, 0, 0, 0, 0, 0}, + []byte{0, 1, 2, 3, 4, 5, 6, 7, 8, 9, 10, 11}, + }, + // Luminance AC. + { + [16]byte{0, 2, 1, 3, 3, 2, 4, 3, 5, 5, 4, 4, 0, 0, 1, 125}, + []byte{ + 0x01, 0x02, 0x03, 0x00, 0x04, 0x11, 0x05, 0x12, + 0x21, 0x31, 0x41, 0x06, 0x13, 0x51, 0x61, 0x07, + 0x22, 0x71, 0x14, 0x32, 0x81, 0x91, 0xa1, 0x08, + 0x23, 0x42, 0xb1, 0xc1, 0x15, 0x52, 0xd1, 0xf0, + 0x24, 0x33, 0x62, 0x72, 0x82, 0x09, 0x0a, 0x16, + 0x17, 0x18, 0x19, 0x1a, 0x25, 0x26, 0x27, 0x28, + 0x29, 0x2a, 0x34, 0x35, 0x36, 0x37, 0x38, 0x39, + 0x3a, 0x43, 0x44, 0x45, 0x46, 0x47, 0x48, 0x49, + 0x4a, 0x53, 0x54, 0x55, 0x56, 0x57, 0x58, 0x59, + 0x5a, 0x63, 0x64, 0x65, 0x66, 0x67, 0x68, 0x69, + 0x6a, 0x73, 0x74, 0x75, 0x76, 0x77, 0x78, 0x79, + 0x7a, 0x83, 0x84, 0x85, 0x86, 0x87, 0x88, 0x89, + 0x8a, 0x92, 0x93, 0x94, 0x95, 0x96, 0x97, 0x98, + 0x99, 0x9a, 0xa2, 0xa3, 0xa4, 0xa5, 0xa6, 0xa7, + 0xa8, 0xa9, 0xaa, 0xb2, 0xb3, 0xb4, 0xb5, 0xb6, + 0xb7, 0xb8, 0xb9, 0xba, 0xc2, 0xc3, 0xc4, 0xc5, + 0xc6, 0xc7, 0xc8, 0xc9, 0xca, 0xd2, 0xd3, 0xd4, + 0xd5, 0xd6, 0xd7, 0xd8, 0xd9, 0xda, 0xe1, 0xe2, + 0xe3, 0xe4, 0xe5, 0xe6, 0xe7, 0xe8, 0xe9, 0xea, + 0xf1, 0xf2, 0xf3, 0xf4, 0xf5, 0xf6, 0xf7, 0xf8, + 0xf9, 0xfa, + }, + }, + // Chrominance DC. + { + [16]byte{0, 3, 1, 1, 1, 1, 1, 1, 1, 1, 1, 0, 0, 0, 0, 0}, + []byte{0, 1, 2, 3, 4, 5, 6, 7, 8, 9, 10, 11}, + }, + // Chrominance AC. + { + [16]byte{0, 2, 1, 2, 4, 4, 3, 4, 7, 5, 4, 4, 0, 1, 2, 119}, + []byte{ + 0x00, 0x01, 0x02, 0x03, 0x11, 0x04, 0x05, 0x21, + 0x31, 0x06, 0x12, 0x41, 0x51, 0x07, 0x61, 0x71, + 0x13, 0x22, 0x32, 0x81, 0x08, 0x14, 0x42, 0x91, + 0xa1, 0xb1, 0xc1, 0x09, 0x23, 0x33, 0x52, 0xf0, + 0x15, 0x62, 0x72, 0xd1, 0x0a, 0x16, 0x24, 0x34, + 0xe1, 0x25, 0xf1, 0x17, 0x18, 0x19, 0x1a, 0x26, + 0x27, 0x28, 0x29, 0x2a, 0x35, 0x36, 0x37, 0x38, + 0x39, 0x3a, 0x43, 0x44, 0x45, 0x46, 0x47, 0x48, + 0x49, 0x4a, 0x53, 0x54, 0x55, 0x56, 0x57, 0x58, + 0x59, 0x5a, 0x63, 0x64, 0x65, 0x66, 0x67, 0x68, + 0x69, 0x6a, 0x73, 0x74, 0x75, 0x76, 0x77, 0x78, + 0x79, 0x7a, 0x82, 0x83, 0x84, 0x85, 0x86, 0x87, + 0x88, 0x89, 0x8a, 0x92, 0x93, 0x94, 0x95, 0x96, + 0x97, 0x98, 0x99, 0x9a, 0xa2, 0xa3, 0xa4, 0xa5, + 0xa6, 0xa7, 0xa8, 0xa9, 0xaa, 0xb2, 0xb3, 0xb4, + 0xb5, 0xb6, 0xb7, 0xb8, 0xb9, 0xba, 0xc2, 0xc3, + 0xc4, 0xc5, 0xc6, 0xc7, 0xc8, 0xc9, 0xca, 0xd2, + 0xd3, 0xd4, 0xd5, 0xd6, 0xd7, 0xd8, 0xd9, 0xda, + 0xe2, 0xe3, 0xe4, 0xe5, 0xe6, 0xe7, 0xe8, 0xe9, + 0xea, 0xf2, 0xf3, 0xf4, 0xf5, 0xf6, 0xf7, 0xf8, + 0xf9, 0xfa, + }, + }, +} + +// huffmanLUT is a compiled look-up table representation of a huffmanSpec. +// Each value maps to a uint32 of which the 8 most significant bits hold the +// codeword size in bits and the 24 least significant bits hold the codeword. +// The maximum codeword size is 16 bits. +type huffmanLUT []uint32 + +func (h *huffmanLUT) init(s huffmanSpec) { + maxValue := 0 + for _, v := range s.value { + if int(v) > maxValue { + maxValue = int(v) + } + } + *h = make([]uint32, maxValue+1) + code, k := uint32(0), 0 + for i := 0; i < len(s.count); i++ { + nBits := uint32(i+1) << 24 + for j := uint8(0); j < s.count[i]; j++ { + (*h)[s.value[k]] = nBits | code + code++ + k++ + } + code <<= 1 + } +} + +// theHuffmanLUT are compiled representations of theHuffmanSpec. +var theHuffmanLUT [4]huffmanLUT + +func init() { + for i, s := range theHuffmanSpec { + theHuffmanLUT[i].init(s) + } +} + +// writer is a buffered writer. +type writer interface { + Flush() os.Error + Write([]byte) (int, os.Error) + WriteByte(byte) os.Error +} + +// encoder encodes an image to the JPEG format. +type encoder struct { + // w is the writer to write to. err is the first error encountered during + // writing. All attempted writes after the first error become no-ops. + w writer + err os.Error + // buf is a scratch buffer. + buf [16]byte + // bits and nBits are accumulated bits to write to w. + bits uint32 + nBits uint8 + // quant is the scaled quantization tables. + quant [nQuantIndex][blockSize]byte +} + +func (e *encoder) flush() { + if e.err != nil { + return + } + e.err = e.w.Flush() +} + +func (e *encoder) write(p []byte) { + if e.err != nil { + return + } + _, e.err = e.w.Write(p) +} + +func (e *encoder) writeByte(b byte) { + if e.err != nil { + return + } + e.err = e.w.WriteByte(b) +} + +// emit emits the least significant nBits bits of bits to the bitstream. +// The precondition is bits < 1<= 8 { + b := uint8(bits >> 24) + e.writeByte(b) + if b == 0xff { + e.writeByte(0x00) + } + bits <<= 8 + nBits -= 8 + } + e.bits, e.nBits = bits, nBits +} + +// emitHuff emits the given value with the given Huffman encoder. +func (e *encoder) emitHuff(h huffIndex, value int) { + x := theHuffmanLUT[h][value] + e.emit(x&(1<<24-1), uint8(x>>24)) +} + +// emitHuffRLE emits a run of runLength copies of value encoded with the given +// Huffman encoder. +func (e *encoder) emitHuffRLE(h huffIndex, runLength, value int) { + a, b := value, value + if a < 0 { + a, b = -value, value-1 + } + var nBits uint8 + if a < 0x100 { + nBits = bitCount[a] + } else { + nBits = 8 + bitCount[a>>8] + } + e.emitHuff(h, runLength<<4|int(nBits)) + if nBits > 0 { + e.emit(uint32(b)&(1<> 8) + e.buf[3] = uint8(markerlen & 0xff) + e.write(e.buf[:4]) +} + +// writeDQT writes the Define Quantization Table marker. +func (e *encoder) writeDQT() { + markerlen := 2 + for _, q := range e.quant { + markerlen += 1 + len(q) + } + e.writeMarkerHeader(dqtMarker, markerlen) + for i, q := range e.quant { + e.writeByte(uint8(i)) + e.write(q[:]) + } +} + +// writeSOF0 writes the Start Of Frame (Baseline) marker. +func (e *encoder) writeSOF0(size image.Point) { + markerlen := 8 + 3*nComponent + e.writeMarkerHeader(sof0Marker, markerlen) + e.buf[0] = 8 // 8-bit color. + e.buf[1] = uint8(size.Y >> 8) + e.buf[2] = uint8(size.Y & 0xff) + e.buf[3] = uint8(size.X >> 8) + e.buf[4] = uint8(size.X & 0xff) + e.buf[5] = nComponent + for i := 0; i < nComponent; i++ { + e.buf[3*i+6] = uint8(i + 1) + // We use 4:2:0 chroma subsampling. + e.buf[3*i+7] = "\x22\x11\x11"[i] + e.buf[3*i+8] = "\x00\x01\x01"[i] + } + e.write(e.buf[:3*(nComponent-1)+9]) +} + +// writeDHT writes the Define Huffman Table marker. +func (e *encoder) writeDHT() { + markerlen := 2 + for _, s := range theHuffmanSpec { + markerlen += 1 + 16 + len(s.value) + } + e.writeMarkerHeader(dhtMarker, markerlen) + for i, s := range theHuffmanSpec { + e.writeByte("\x00\x10\x01\x11"[i]) + e.write(s.count[:]) + e.write(s.value) + } +} + +// writeBlock writes a block of pixel data using the given quantization table, +// returning the post-quantized DC value of the DCT-transformed block. +func (e *encoder) writeBlock(b *block, q quantIndex, prevDC int) int { + fdct(b) + // Emit the DC delta. + dc := div(b[0], (8 * int(e.quant[q][0]))) + e.emitHuffRLE(huffIndex(2*q+0), 0, dc-prevDC) + // Emit the AC components. + h, runLength := huffIndex(2*q+1), 0 + for k := 1; k < blockSize; k++ { + ac := div(b[unzig[k]], (8 * int(e.quant[q][k]))) + if ac == 0 { + runLength++ + } else { + for runLength > 15 { + e.emitHuff(h, 0xf0) + runLength -= 16 + } + e.emitHuffRLE(h, runLength, ac) + runLength = 0 + } + } + if runLength > 0 { + e.emitHuff(h, 0x00) + } + return dc +} + +// toYCbCr converts the 8x8 region of m whose top-left corner is p to its +// YCbCr values. +func toYCbCr(m image.Image, p image.Point, yBlock, cbBlock, crBlock *block) { + b := m.Bounds() + xmax := b.Max.X - 1 + ymax := b.Max.Y - 1 + for j := 0; j < 8; j++ { + for i := 0; i < 8; i++ { + r, g, b, _ := m.At(min(p.X+i, xmax), min(p.Y+j, ymax)).RGBA() + yy, cb, cr := ycbcr.RGBToYCbCr(uint8(r>>8), uint8(g>>8), uint8(b>>8)) + yBlock[8*j+i] = int(yy) + cbBlock[8*j+i] = int(cb) + crBlock[8*j+i] = int(cr) + } + } +} + +// scale scales the 16x16 region represented by the 4 src blocks to the 8x8 +// dst block. +func scale(dst *block, src *[4]block) { + for i := 0; i < 4; i++ { + dstOff := (i&2)<<4 | (i&1)<<2 + for y := 0; y < 4; y++ { + for x := 0; x < 4; x++ { + j := 16*y + 2*x + sum := src[i][j] + src[i][j+1] + src[i][j+8] + src[i][j+9] + dst[8*y+x+dstOff] = (sum + 2) >> 2 + } + } + } +} + +// sosHeader is the SOS marker "\xff\xda" followed by 12 bytes: +// - the marker length "\x00\x0c", +// - the number of components "\x03", +// - component 1 uses DC table 0 and AC table 0 "\x01\x00", +// - component 2 uses DC table 1 and AC table 1 "\x02\x11", +// - component 3 uses DC table 1 and AC table 1 "\x03\x11", +// - padding "\x00\x00\x00". +var sosHeader = []byte{ + 0xff, 0xda, 0x00, 0x0c, 0x03, 0x01, 0x00, 0x02, + 0x11, 0x03, 0x11, 0x00, 0x00, 0x00, +} + +// writeSOS writes the StartOfScan marker. +func (e *encoder) writeSOS(m image.Image) { + e.write(sosHeader) + var ( + // Scratch buffers to hold the YCbCr values. + yBlock block + cbBlock [4]block + crBlock [4]block + cBlock block + // DC components are delta-encoded. + prevDCY, prevDCCb, prevDCCr int + ) + bounds := m.Bounds() + for y := bounds.Min.Y; y < bounds.Max.Y; y += 16 { + for x := bounds.Min.X; x < bounds.Max.X; x += 16 { + for i := 0; i < 4; i++ { + xOff := (i & 1) * 8 + yOff := (i & 2) * 4 + p := image.Point{x + xOff, y + yOff} + toYCbCr(m, p, &yBlock, &cbBlock[i], &crBlock[i]) + prevDCY = e.writeBlock(&yBlock, 0, prevDCY) + } + scale(&cBlock, &cbBlock) + prevDCCb = e.writeBlock(&cBlock, 1, prevDCCb) + scale(&cBlock, &crBlock) + prevDCCr = e.writeBlock(&cBlock, 1, prevDCCr) + } + } + // Pad the last byte with 1's. + e.emit(0x7f, 7) +} + +// DefaultQuality is the default quality encoding parameter. +const DefaultQuality = 75 + +// Options are the encoding parameters. +// Quality ranges from 1 to 100 inclusive, higher is better. +type Options struct { + Quality int +} + +// Encode writes the Image m to w in JPEG 4:2:0 baseline format with the given +// options. Default parameters are used if a nil *Options is passed. +func Encode(w io.Writer, m image.Image, o *Options) os.Error { + b := m.Bounds() + if b.Dx() >= 1<<16 || b.Dy() >= 1<<16 { + return os.NewError("jpeg: image is too large to encode") + } + var e encoder + if ww, ok := w.(writer); ok { + e.w = ww + } else { + e.w = bufio.NewWriter(w) + } + // Clip quality to [1, 100]. + quality := DefaultQuality + if o != nil { + quality = o.Quality + if quality < 1 { + quality = 1 + } else if quality > 100 { + quality = 100 + } + } + // Convert from a quality rating to a scaling factor. + var scale int + if quality < 50 { + scale = 5000 / quality + } else { + scale = 200 - quality*2 + } + // Initialize the quantization tables. + for i := range e.quant { + for j := range e.quant[i] { + x := int(unscaledQuant[i][j]) + x = (x*scale + 50) / 100 + if x < 1 { + x = 1 + } else if x > 255 { + x = 255 + } + e.quant[i][j] = uint8(x) + } + } + // Write the Start Of Image marker. + e.buf[0] = 0xff + e.buf[1] = 0xd8 + e.write(e.buf[:2]) + // Write the quantization tables. + e.writeDQT() + // Write the image dimensions. + e.writeSOF0(b.Size()) + // Write the Huffman tables. + e.writeDHT() + // Write the image data. + e.writeSOS(m) + // Write the End Of Image marker. + e.buf[0] = 0xff + e.buf[1] = 0xd9 + e.write(e.buf[:2]) + e.flush() + return e.err +} diff --git a/src/pkg/image/jpeg/writer_test.go b/src/pkg/image/jpeg/writer_test.go new file mode 100644 index 0000000000..00922dd5ce --- /dev/null +++ b/src/pkg/image/jpeg/writer_test.go @@ -0,0 +1,87 @@ +// Copyright 2011 The Go Authors. All rights reserved. +// Use of this source code is governed by a BSD-style +// license that can be found in the LICENSE file. + +package jpeg + +import ( + "bytes" + "image" + "image/png" + "os" + "testing" +) + +var testCase = []struct { + filename string + quality int + tolerance int64 +}{ + {"../testdata/video-001.png", 1, 24 << 8}, + {"../testdata/video-001.png", 20, 12 << 8}, + {"../testdata/video-001.png", 60, 8 << 8}, + {"../testdata/video-001.png", 80, 6 << 8}, + {"../testdata/video-001.png", 90, 4 << 8}, + {"../testdata/video-001.png", 100, 2 << 8}, +} + +func delta(u0, u1 uint32) int64 { + d := int64(u0) - int64(u1) + if d < 0 { + return -d + } + return d +} + +func readPng(filename string) (image.Image, os.Error) { + f, err := os.Open(filename) + if err != nil { + return nil, err + } + defer f.Close() + return png.Decode(f) +} + +func TestWriter(t *testing.T) { + for _, tc := range testCase { + // Read the image. + m0, err := readPng(tc.filename) + if err != nil { + t.Error(tc.filename, err) + continue + } + // Encode that image as JPEG. + buf := bytes.NewBuffer(nil) + err = Encode(buf, m0, &Options{Quality: tc.quality}) + if err != nil { + t.Error(tc.filename, err) + continue + } + // Decode that JPEG. + m1, err := Decode(buf) + if err != nil { + t.Error(tc.filename, err) + continue + } + // Compute the average delta in RGB space. + b := m0.Bounds() + var sum, n int64 + for y := b.Min.Y; y < b.Max.Y; y++ { + for x := b.Min.X; x < b.Max.X; x++ { + c0 := m0.At(x, y) + c1 := m1.At(x, y) + r0, g0, b0, _ := c0.RGBA() + r1, g1, b1, _ := c1.RGBA() + sum += delta(r0, r1) + sum += delta(g0, g1) + sum += delta(b0, b1) + n += 3 + } + } + // Compare the average delta to the tolerance level. + if sum/n > tc.tolerance { + t.Errorf("%s, quality=%d: average delta is too high", tc.filename, tc.quality) + continue + } + } +}