/* * jquant2.c * * This file was part of the Independent JPEG Group's software: * Copyright (C) 1991-1996, Thomas G. Lane. * libjpeg-turbo Modifications: * Copyright (C) 2009, 2014-2015, 2020, D. R. Commander. * For conditions of distribution and use, see the accompanying README.ijg * file. * * This file contains 2-pass color quantization (color mapping) routines. * These routines provide selection of a custom color map for an image, * followed by mapping of the image to that color map, with optional * Floyd-Steinberg dithering. * It is also possible to use just the second pass to map to an arbitrary * externally-given color map. * * Note: ordered dithering is not supported, since there isn't any fast * way to compute intercolor distances; it's unclear that ordered dither's * fundamental assumptions even hold with an irregularly spaced color map. */ #define JPEG_INTERNALS #include "jinclude.h" #include "jpeglib.h" #ifdef QUANT_2PASS_SUPPORTED /* * This module implements the well-known Heckbert paradigm for color * quantization. Most of the ideas used here can be traced back to * Heckbert's seminal paper * Heckbert, Paul. "Color Image Quantization for Frame Buffer Display", * Proc. SIGGRAPH '82, Computer Graphics v.16 #3 (July 1982), pp 297-304. * * In the first pass over the image, we accumulate a histogram showing the * usage count of each possible color. To keep the histogram to a reasonable * size, we reduce the precision of the input; typical practice is to retain * 5 or 6 bits per color, so that 8 or 4 different input values are counted * in the same histogram cell. * * Next, the color-selection step begins with a box representing the whole * color space, and repeatedly splits the "largest" remaining box until we * have as many boxes as desired colors. Then the mean color in each * remaining box becomes one of the possible output colors. * * The second pass over the image maps each input pixel to the closest output * color (optionally after applying a Floyd-Steinberg dithering correction). * This mapping is logically trivial, but making it go fast enough requires * considerable care. * * Heckbert-style quantizers vary a good deal in their policies for choosing * the "largest" box and deciding where to cut it. The particular policies * used here have proved out well in experimental comparisons, but better ones * may yet be found. * * In earlier versions of the IJG code, this module quantized in YCbCr color * space, processing the raw upsampled data without a color conversion step. * This allowed the color conversion math to be done only once per colormap * entry, not once per pixel. However, that optimization precluded other * useful optimizations (such as merging color conversion with upsampling) * and it also interfered with desired capabilities such as quantizing to an * externally-supplied colormap. We have therefore abandoned that approach. * The present code works in the post-conversion color space, typically RGB. * * To improve the visual quality of the results, we actually work in scaled * RGB space, giving G distances more weight than R, and R in turn more than * B. To do everything in integer math, we must use integer scale factors. * The 2/3/1 scale factors used here correspond loosely to the relative * weights of the colors in the NTSC grayscale equation. * If you want to use this code to quantize a non-RGB color space, you'll * probably need to change these scale factors. */ #define R_SCALE … #define G_SCALE … #define B_SCALE … static const int c_scales[3] = …; #define C0_SCALE … #define C1_SCALE … #define C2_SCALE … /* * First we have the histogram data structure and routines for creating it. * * The number of bits of precision can be adjusted by changing these symbols. * We recommend keeping 6 bits for G and 5 each for R and B. * If you have plenty of memory and cycles, 6 bits all around gives marginally * better results; if you are short of memory, 5 bits all around will save * some space but degrade the results. * To maintain a fully accurate histogram, we'd need to allocate a "long" * (preferably unsigned long) for each cell. In practice this is overkill; * we can get by with 16 bits per cell. Few of the cell counts will overflow, * and clamping those that do overflow to the maximum value will give close- * enough results. This reduces the recommended histogram size from 256Kb * to 128Kb, which is a useful savings on PC-class machines. * (In the second pass the histogram space is re-used for pixel mapping data; * in that capacity, each cell must be able to store zero to the number of * desired colors. 16 bits/cell is plenty for that too.) * Since the JPEG code is intended to run in small memory model on 80x86 * machines, we can't just allocate the histogram in one chunk. Instead * of a true 3-D array, we use a row of pointers to 2-D arrays. Each * pointer corresponds to a C0 value (typically 2^5 = 32 pointers) and * each 2-D array has 2^6*2^5 = 2048 or 2^6*2^6 = 4096 entries. */ #define MAXNUMCOLORS … /* These will do the right thing for either R,G,B or B,G,R color order, * but you may not like the results for other color orders. */ #define HIST_C0_BITS … #define HIST_C1_BITS … #define HIST_C2_BITS … /* Number of elements along histogram axes. */ #define HIST_C0_ELEMS … #define HIST_C1_ELEMS … #define HIST_C2_ELEMS … /* These are the amounts to shift an input value to get a histogram index. */ #define C0_SHIFT … #define C1_SHIFT … #define C2_SHIFT … histcell; /* histogram cell; prefer an unsigned type */ histptr; /* for pointers to histogram cells */ hist1d; /* typedefs for the array */ hist2d; /* type for the 2nd-level pointers */ hist3d; /* type for top-level pointer */ /* Declarations for Floyd-Steinberg dithering. * * Errors are accumulated into the array fserrors[], at a resolution of * 1/16th of a pixel count. The error at a given pixel is propagated * to its not-yet-processed neighbors using the standard F-S fractions, * ... (here) 7/16 * 3/16 5/16 1/16 * We work left-to-right on even rows, right-to-left on odd rows. * * We can get away with a single array (holding one row's worth of errors) * by using it to store the current row's errors at pixel columns not yet * processed, but the next row's errors at columns already processed. We * need only a few extra variables to hold the errors immediately around the * current column. (If we are lucky, those variables are in registers, but * even if not, they're probably cheaper to access than array elements are.) * * The fserrors[] array has (#columns + 2) entries; the extra entry at * each end saves us from special-casing the first and last pixels. * Each entry is three values long, one value for each color component. */ #if BITS_IN_JSAMPLE == 8 FSERROR; /* 16 bits should be enough */ LOCFSERROR; /* use 'int' for calculation temps */ #else typedef JLONG FSERROR; /* may need more than 16 bits */ typedef JLONG LOCFSERROR; /* be sure calculation temps are big enough */ #endif FSERRPTR; /* pointer to error array */ /* Private subobject */ my_cquantizer; my_cquantize_ptr; /* * Prescan some rows of pixels. * In this module the prescan simply updates the histogram, which has been * initialized to zeroes by start_pass. * An output_buf parameter is required by the method signature, but no data * is actually output (in fact the buffer controller is probably passing a * NULL pointer). */ METHODDEF(void) prescan_quantize(j_decompress_ptr cinfo, JSAMPARRAY input_buf, JSAMPARRAY output_buf, int num_rows) { … } /* * Next we have the really interesting routines: selection of a colormap * given the completed histogram. * These routines work with a list of "boxes", each representing a rectangular * subset of the input color space (to histogram precision). */ box; boxptr; LOCAL(boxptr) find_biggest_color_pop(boxptr boxlist, int numboxes) /* Find the splittable box with the largest color population */ /* Returns NULL if no splittable boxes remain */ { … } LOCAL(boxptr) find_biggest_volume(boxptr boxlist, int numboxes) /* Find the splittable box with the largest (scaled) volume */ /* Returns NULL if no splittable boxes remain */ { … } LOCAL(void) update_box(j_decompress_ptr cinfo, boxptr boxp) /* Shrink the min/max bounds of a box to enclose only nonzero elements, */ /* and recompute its volume and population */ { … } LOCAL(int) median_cut(j_decompress_ptr cinfo, boxptr boxlist, int numboxes, int desired_colors) /* Repeatedly select and split the largest box until we have enough boxes */ { … } LOCAL(void) compute_color(j_decompress_ptr cinfo, boxptr boxp, int icolor) /* Compute representative color for a box, put it in colormap[icolor] */ { … } LOCAL(void) select_colors(j_decompress_ptr cinfo, int desired_colors) /* Master routine for color selection */ { … } /* * These routines are concerned with the time-critical task of mapping input * colors to the nearest color in the selected colormap. * * We re-use the histogram space as an "inverse color map", essentially a * cache for the results of nearest-color searches. All colors within a * histogram cell will be mapped to the same colormap entry, namely the one * closest to the cell's center. This may not be quite the closest entry to * the actual input color, but it's almost as good. A zero in the cache * indicates we haven't found the nearest color for that cell yet; the array * is cleared to zeroes before starting the mapping pass. When we find the * nearest color for a cell, its colormap index plus one is recorded in the * cache for future use. The pass2 scanning routines call fill_inverse_cmap * when they need to use an unfilled entry in the cache. * * Our method of efficiently finding nearest colors is based on the "locally * sorted search" idea described by Heckbert and on the incremental distance * calculation described by Spencer W. Thomas in chapter III.1 of Graphics * Gems II (James Arvo, ed. Academic Press, 1991). Thomas points out that * the distances from a given colormap entry to each cell of the histogram can * be computed quickly using an incremental method: the differences between * distances to adjacent cells themselves differ by a constant. This allows a * fairly fast implementation of the "brute force" approach of computing the * distance from every colormap entry to every histogram cell. Unfortunately, * it needs a work array to hold the best-distance-so-far for each histogram * cell (because the inner loop has to be over cells, not colormap entries). * The work array elements have to be JLONGs, so the work array would need * 256Kb at our recommended precision. This is not feasible in DOS machines. * * To get around these problems, we apply Thomas' method to compute the * nearest colors for only the cells within a small subbox of the histogram. * The work array need be only as big as the subbox, so the memory usage * problem is solved. Furthermore, we need not fill subboxes that are never * referenced in pass2; many images use only part of the color gamut, so a * fair amount of work is saved. An additional advantage of this * approach is that we can apply Heckbert's locality criterion to quickly * eliminate colormap entries that are far away from the subbox; typically * three-fourths of the colormap entries are rejected by Heckbert's criterion, * and we need not compute their distances to individual cells in the subbox. * The speed of this approach is heavily influenced by the subbox size: too * small means too much overhead, too big loses because Heckbert's criterion * can't eliminate as many colormap entries. Empirically the best subbox * size seems to be about 1/512th of the histogram (1/8th in each direction). * * Thomas' article also describes a refined method which is asymptotically * faster than the brute-force method, but it is also far more complex and * cannot efficiently be applied to small subboxes. It is therefore not * useful for programs intended to be portable to DOS machines. On machines * with plenty of memory, filling the whole histogram in one shot with Thomas' * refined method might be faster than the present code --- but then again, * it might not be any faster, and it's certainly more complicated. */ /* log2(histogram cells in update box) for each axis; this can be adjusted */ #define BOX_C0_LOG … #define BOX_C1_LOG … #define BOX_C2_LOG … #define BOX_C0_ELEMS … #define BOX_C1_ELEMS … #define BOX_C2_ELEMS … #define BOX_C0_SHIFT … #define BOX_C1_SHIFT … #define BOX_C2_SHIFT … /* * The next three routines implement inverse colormap filling. They could * all be folded into one big routine, but splitting them up this way saves * some stack space (the mindist[] and bestdist[] arrays need not coexist) * and may allow some compilers to produce better code by registerizing more * inner-loop variables. */ LOCAL(int) find_nearby_colors(j_decompress_ptr cinfo, int minc0, int minc1, int minc2, JSAMPLE colorlist[]) /* Locate the colormap entries close enough to an update box to be candidates * for the nearest entry to some cell(s) in the update box. The update box * is specified by the center coordinates of its first cell. The number of * candidate colormap entries is returned, and their colormap indexes are * placed in colorlist[]. * This routine uses Heckbert's "locally sorted search" criterion to select * the colors that need further consideration. */ { … } LOCAL(void) find_best_colors(j_decompress_ptr cinfo, int minc0, int minc1, int minc2, int numcolors, JSAMPLE colorlist[], JSAMPLE bestcolor[]) /* Find the closest colormap entry for each cell in the update box, * given the list of candidate colors prepared by find_nearby_colors. * Return the indexes of the closest entries in the bestcolor[] array. * This routine uses Thomas' incremental distance calculation method to * find the distance from a colormap entry to successive cells in the box. */ { … } LOCAL(void) fill_inverse_cmap(j_decompress_ptr cinfo, int c0, int c1, int c2) /* Fill the inverse-colormap entries in the update box that contains */ /* histogram cell c0/c1/c2. (Only that one cell MUST be filled, but */ /* we can fill as many others as we wish.) */ { … } /* * Map some rows of pixels to the output colormapped representation. */ METHODDEF(void) pass2_no_dither(j_decompress_ptr cinfo, JSAMPARRAY input_buf, JSAMPARRAY output_buf, int num_rows) /* This version performs no dithering */ { … } METHODDEF(void) pass2_fs_dither(j_decompress_ptr cinfo, JSAMPARRAY input_buf, JSAMPARRAY output_buf, int num_rows) /* This version performs Floyd-Steinberg dithering */ { … } /* * Initialize the error-limiting transfer function (lookup table). * The raw F-S error computation can potentially compute error values of up to * +- MAXJSAMPLE. But we want the maximum correction applied to a pixel to be * much less, otherwise obviously wrong pixels will be created. (Typical * effects include weird fringes at color-area boundaries, isolated bright * pixels in a dark area, etc.) The standard advice for avoiding this problem * is to ensure that the "corners" of the color cube are allocated as output * colors; then repeated errors in the same direction cannot cause cascading * error buildup. However, that only prevents the error from getting * completely out of hand; Aaron Giles reports that error limiting improves * the results even with corner colors allocated. * A simple clamping of the error values to about +- MAXJSAMPLE/8 works pretty * well, but the smoother transfer function used below is even better. Thanks * to Aaron Giles for this idea. */ LOCAL(void) init_error_limit(j_decompress_ptr cinfo) /* Allocate and fill in the error_limiter table */ { … } /* * Finish up at the end of each pass. */ METHODDEF(void) finish_pass1(j_decompress_ptr cinfo) { … } METHODDEF(void) finish_pass2(j_decompress_ptr cinfo) { … } /* * Initialize for each processing pass. */ METHODDEF(void) start_pass_2_quant(j_decompress_ptr cinfo, boolean is_pre_scan) { … } /* * Switch to a new external colormap between output passes. */ METHODDEF(void) new_color_map_2_quant(j_decompress_ptr cinfo) { … } /* * Module initialization routine for 2-pass color quantization. */ GLOBAL(void) jinit_2pass_quantizer(j_decompress_ptr cinfo) { … } #endif /* QUANT_2PASS_SUPPORTED */
Codebase Browser
chromium