...
Run Format

Source file src/image/jpeg/fdct.go

Documentation: image/jpeg

  // 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)
  	}
  }
  

View as plain text