写selenium自动化时，为了查看运行效果，后给浏览器截图，想到可以生成gif图片来快速预览。看到已经有人实现了，直接拿过来。作者是Kevin Weiner。
共涉及到三个java文件，分别是NeuQuant.java,LZWEncoder.java,AnimatedGifEncoder.java,有了这三个文件，我们可以自己编写方法调用，
代码如下：
BufferedImage src1 = ImageIO.read(new File("Img221785570.jpg"));
BufferedImage src2 = ImageIO.read(new File("W.gif"));
//BufferedImage src3 = ImageIO.read(new File("c:/ship3.jpg"));
AnimatedGifEncoder e = new AnimatedGifEncoder();
e.setRepeat(0);
e.start("laoma.gif");
e.setDelay(300); // 1 frame per sec
e.addFrame(src1);
e.setDelay(100);
e.addFrame(src2);
e.setDelay(100);
// e.addFrame(src2);
e.finish();
三个java文件源码如下：
NeuQuant.java
public class NeuQuant
{
protected static final int netsize = 256; /* number of colours used */
/* four primes near 500 - assume no image has a length so large */
/* that it is divisible by all four primes */
protected static final int prime1 = 499;
protected static final int prime2 = 491;
protected static final int prime3 = 487;
protected static final int prime4 = 503;
protected static final int minpicturebytes = (3 * prime4);
/* minimum size for input image */
/* Program Skeleton
----------------
[select samplefac in range 1..30]
[read image from input file]
pic = (unsigned char*) malloc(3*width*height);
initnet(pic,3*width*height,samplefac);
learn();
unbiasnet();
[write output image header, using writecolourmap(f)]
inxbuild();
write output image using inxsearch(b,g,r) */
/* Network Definitions
------------------- */
protected static final int maxnetpos = (netsize - 1);
protected static final int netbiasshift = 4; /* bias for colour values */
protected static final int ncycles = 100; /* no. of learning cycles */
/* defs for freq and bias */
protected static final int intbiasshift = 16; /* bias for fractions */
protected static final int intbias = (((int) 1) << intbiasshift);
protected static final int gammashift = 10; /* gamma = 1024 */
protected static final int gamma = (((int) 1) << gammashift);
protected static final int betashift = 10;
protected static final int beta = (intbias >> betashift); /* beta = 1/1024 */
protected static final int betagamma =
(intbias << (gammashift - betashift));
/* defs for decreasing radius factor */
protected static final int initrad = (netsize >> 3); /* for 256 cols, radius starts */
protected static final int radiusbiasshift = 6; /* at 32.0 biased by 6 bits */
protected static final int radiusbias = (((int) 1) << radiusbiasshift);
protected static final int initradius = (initrad * radiusbias); /* and decreases by a */
protected static final int radiusdec = 30; /* factor of 1/30 each cycle */
/* defs for decreasing alpha factor */
protected static final int alphabiasshift = 10; /* alpha starts at 1.0 */
protected static final int initalpha = (((int) 1) << alphabiasshift);
protected int alphadec; /* biased by 10 bits */
/* radbias and alpharadbias used for radpower calculation */
protected static final int radbiasshift = 8;
protected static final int radbias = (((int) 1) << radbiasshift);
protected static final int alpharadbshift = (alphabiasshift + radbiasshift);
protected static final int alpharadbias = (((int) 1) << alpharadbshift);
/* Types and Global Variables
-------------------------- */
protected byte[] thepicture; /* the input image itself */
protected int lengthcount; /* lengthcount = H*W*3 */
protected int samplefac; /* sampling factor 1..30 */
// typedef int pixel[4]; /* BGRc */
protected int[][] network; /* the network itself - [netsize][4] */
protected int[] netindex = new int[256];
/* for network lookup - really 256 */
protected int[] bias = new int[netsize];
/* bias and freq arrays for learning */
protected int[] freq = new int[netsize];
protected int[] radpower = new int[initrad];
/* radpower for precomputation */
/* Initialise network in range (0,0,0) to (255,255,255) and set parameters
----------------------------------------------------------------------- */
public NeuQuant(byte[] thepic, int len, int sample) {
int i;
int[] p;
thepicture = thepic;
lengthcount = len;
samplefac = sample;
network = new int[netsize][];
for (i = 0; i < netsize; i++) {
network[i] = new int[4];
p = network[i];
p[0] = p[1] = p[2] = (i << (netbiasshift + 8)) / netsize;
freq[i] = intbias / netsize; /* 1/netsize */
bias[i] = 0;
}
}
public byte[] colorMap() {
byte[] map = new byte[3 * netsize];
int[] index = new int[netsize];
for (int i = 0; i < netsize; i++)
index[network[i][3]] = i;
int k = 0;
for (int i = 0; i < netsize; i++) {
int j = index[i];
map[k++] = (byte) (network[j][0]);
map[k++] = (byte) (network[j][1]);
map[k++] = (byte) (network[j][2]);
}
return map;
}
/* Insertion sort of network and building of netindex[0..255] (to do after unbias)
------------------------------------------------------------------------------- */
public void inxbuild() {
int i, j, smallpos, smallval;
int[] p;
int[] q;
int previouscol, startpos;
previouscol = 0;
startpos = 0;
for (i = 0; i < netsize; i++) {
p = network[i];
smallpos = i;
smallval = p[1]; /* index on g */
/* find smallest in i..netsize-1 */
for (j = i + 1; j < netsize; j++) {
q = network[j];
if (q[1] < smallval) { /* index on g */
smallpos = j;
smallval = q[1]; /* index on g */
}
}
q = network[smallpos];
/* swap p (i) and q (smallpos) entries */
if (i != smallpos) {
j = q[0];
q[0] = p[0];
p[0] = j;
j = q[1];
q[1] = p[1];
p[1] = j;
j = q[2];
q[2] = p[2];
p[2] = j;
j = q[3];
q[3] = p[3];
p[3] = j;
}
/* smallval entry is now in position i */
if (smallval != previouscol) {
netindex[previouscol] = (startpos + i) >> 1;
for (j = previouscol + 1; j < smallval; j++)
netindex[j] = i;
previouscol = smallval;
startpos = i;
}
}
netindex[previouscol] = (startpos + maxnetpos) >> 1;
for (j = previouscol + 1; j < 256; j++)
netindex[j] = maxnetpos; /* really 256 */
}
/* Main Learning Loop
------------------ */
public void learn() {
int i, j, b, g, r;
int radius, rad, alpha, step, delta, samplepixels;
byte[] p;
int pix, lim;
if (lengthcount < minpicturebytes)
samplefac = 1;
alphadec = 30 + ((samplefac - 1) / 3);
p = thepicture;
pix = 0;
lim = lengthcount;
samplepixels = lengthcount / (3 * samplefac);
delta = samplepixels / ncycles;
alpha = initalpha;
radius = initradius;
rad = radius >> radiusbiasshift;
if (rad <= 1)
rad = 0;
for (i = 0; i < rad; i++)
radpower[i] =
alpha * (((rad * rad - i * i) * radbias) / (rad * rad));
//fprintf(stderr,"beginning 1D learning: initial radius=%d/n", rad);
if (lengthcount < minpicturebytes)
step = 3;
else if ((lengthcount % prime1) != 0)
step = 3 * prime1;
else {
if ((lengthcount % prime2) != 0)
step = 3 * prime2;
else {
if ((lengthcount % prime3) != 0)
step = 3 * prime3;
else
step = 3 * prime4;
}
}
i = 0;
while (i < samplepixels) {
b = (p[pix + 0] & 0xff) << netbiasshift;
g = (p[pix + 1] & 0xff) << netbiasshift;
r = (p[pix + 2] & 0xff) << netbiasshift;
j = contest(b, g, r);
altersingle(alpha, j, b, g, r);
if (rad != 0)
alterneigh(rad, j, b, g, r); /* alter neighbours */
pix += step;
if (pix >= lim)
pix -= lengthcount;
i++;
if (delta == 0)
delta = 1;
if (i % delta == 0) {
alpha -= alpha / alphadec;
radius -= radius / radiusdec;
rad = radius >> radiusbiasshift;
if (rad <= 1)
rad = 0;
for (j = 0; j < rad; j++)
radpower[j] =
alpha * (((rad * rad - j * j) * radbias) / (rad * rad));
}
}
//fprintf(stderr,"finished 1D learning: final alpha=%f !/n",((float)alpha)/initalpha);
}
/* Search for BGR values 0..255 (after net is unbiased) and return colour index
---------------------------------------------------------------------------- */
public int map(int b, int g, int r) {
int i, j, dist, a, bestd;
int[] p;
int best;
bestd = 1000; /* biggest possible dist is 256*3 */
best = -1;
i = netindex[g]; /* index on g */
j = i - 1; /* start at netindex[g] and work outwards */
while ((i < netsize) || (j >= 0)) {
if (i < netsize) {
p = network[i];
dist = p[1] - g; /* inx key */
if (dist >= bestd)
i = netsize; /* stop iter */
else {
i++;
if (dist < 0)
dist = -dist;
a = p[0] - b;
if (a < 0)
a = -a;
dist += a;
if (dist < bestd) {
a = p[2] - r;
if (a < 0)
a = -a;
dist += a;
if (dist < bestd) {
bestd = dist;
best = p[3];
}
}
}
}
if (j >= 0) {
p = network[j];
dist = g - p[1]; /* inx key - reverse dif */
if (dist >= bestd)
j = -1; /* stop iter */
else {
j--;
if (dist < 0)
dist = -dist;
a = p[0] - b;
if (a < 0)
a = -a;
dist += a;
if (dist < bestd) {
a = p[2] - r;
if (a < 0)
a = -a;
dist += a;
if (dist < bestd) {
bestd = dist;
best = p[3];
}
}
}
}
}
return (best);
}
public byte[] process() {
learn();
unbiasnet();
inxbuild();
return colorMap();
}
/* Unbias network to give byte values 0..255 and record position i to prepare for sort
----------------------------------------------------------------------------------- */
public void unbiasnet() {
int i, j;
for (i = 0; i < netsize; i++) {
network[i][0] >>= netbiasshift;
network[i][1] >>= netbiasshift;
network[i][2] >>= netbiasshift;
network[i][3] = i; /* record colour no */
}
}
/* Move adjacent neurons by precomputed alpha*(1-((i-j)^2/[r]^2)) in radpower[|i-j|]
--------------------------------------------------------------------------------- */
protected void alterneigh(int rad, int i, int b, int g, int r) {
int j, k, lo, hi, a, m;
int[] p;
lo = i - rad;
if (lo < -1)
lo = -1;
hi = i + rad;
if (hi > netsize)
hi = netsize;
j = i + 1;
k = i - 1;
m = 1;
while ((j < hi) || (k > lo)) {
a = radpower[m++];
if (j < hi) {
p = network[j++];
try {
p[0] -= (a * (p[0] - b)) / alpharadbias;
p[1] -= (a * (p[1] - g)) / alpharadbias;
p[2] -= (a * (p[2] - r)) / alpharadbias;
} catch (Exception e) {
} // prevents 1.3 miscompilation
}
if (k > lo) {
p = network[k--];
try {
p[0] -= (a * (p[0] - b)) / alpharadbias;
p[1] -= (a * (p[1] - g)) / alpharadbias;
p[2] -= (a * (p[2] - r)) / alpharadbias;
} catch (Exception e) {
}
}
}
}
/* Move neuron i towards biased (b,g,r) by factor alpha
---------------------------------------------------- */
protected void altersingle(int alpha, int i, int b, int g, int r) {
/* alter hit neuron */
int[] n = network[i];
n[0] -= (alpha * (n[0] - b)) / initalpha;
n[1] -= (alpha * (n[1] - g)) / initalpha;
n[2] -= (alpha * (n[2] - r)) / initalpha;
}
/* Search for biased BGR values
---------------------------- */
protected int contest(int b, int g, int r) {
/* finds closest neuron (min dist) and updates freq */
/* finds best neuron (min dist-bias) and returns position */
/* for frequently chosen neurons, freq[i] is high and bias[i] is negative */
/* bias[i] = gamma*((1/netsize)-freq[i]) */
int i, dist, a, biasdist, betafreq;
int bestpos, bestbiaspos, bestd, bestbiasd;
int[] n;
bestd = ~(((int) 1) << 31);
bestbiasd = bestd;
bestpos = -1;
bestbiaspos = bestpos;
for (i = 0; i < netsize; i++) {
n = network[i];
dist = n[0] - b;
if (dist < 0)
dist = -dist;
a = n[1] - g;
if (a < 0)
a = -a;
dist += a;
a = n[2] - r;
if (a < 0)
a = -a;
dist += a;
if (dist < bestd) {
bestd = dist;
bestpos = i;
}
biasdist = dist - ((bias[i]) >> (intbiasshift - netbiasshift));
if (biasdist < bestbiasd) {
bestbiasd = biasdist;
bestbiaspos = i;
}
betafreq = (freq[i] >> betashift);
freq[i] -= betafreq;
bias[i] += (betafreq << gammashift);
}
freq[bestpos] += beta;
bias[bestpos] -= betagamma;
return (bestbiaspos);
}
}
LZWEncoder.java源码如下：
package com.yeetrack.selenium.Image;
import java.io.OutputStream;
import java.io.IOException;
//==============================================================================
// Adapted from Jef Poskanzer‘s Java port by way of J. M. G. Elliott.
// K Weiner 12/00
class LZWEncoder {
private static final int EOF = -1;
private int imgW, imgH;
private byte[] pixAry;
private int initCodeSize;
private int remaining;
private int curPixel;
// GIFCOMPR.C - GIF Image compression routines
//
// Lempel-Ziv compression based on ‘compress‘. GIF modifications by
// David Rowley (mgardi@watdcsu.waterloo.edu)
// General DEFINEs
static final int BITS = 12;
static final int HSIZE = 5003; // 80% occupancy
// GIF Image compression - modified ‘compress‘
//
// Based on: compress.c - File compression ala IEEE Computer, June 1984.
//
// By Authors: Spencer W. Thomas (decvax!harpo!utah-cs!utah-gr!thomas)
// Jim McKie (decvax!mcvax!jim)
// Steve Davies (decvax!vax135!petsd!peora!srd)
// Ken Turkowski (decvax!decwrl!turtlevax!ken)
// James A. Woods (decvax!ihnp4!ames!jaw)
// Joe Orost (decvax!vax135!petsd!joe)
int n_bits; // number of bits/code
int maxbits = BITS; // user settable max # bits/code
int maxcode; // maximum code, given n_bits
int maxmaxcode = 1 << BITS; // should NEVER generate this code
int[] htab = new int[HSIZE];
int[] codetab = new int[HSIZE];
int hsize = HSIZE; // for dynamic table sizing
int free_ent = 0; // first unused entry
// block compression parameters -- after all codes are used up,
// and compression rate changes, start over.
boolean clear_flg = false;
// Algorithm: use open addressing double hashing (no chaining) on the
// prefix code / next character combination. We do a variant of Knuth‘s
// algorithm D (vol. 3, sec. 6.4) along with G. Knott‘s relatively-prime
// secondary probe. Here, the modular division first probe is gives way
// to a faster exclusive-or manipulation. Also do block compression with
// an adaptive reset, whereby the code table is cleared when the compression
// ratio decreases, but after the table fills. The variable-length output
// codes are re-sized at this point, and a special CLEAR code is generated
// for the decompressor. Late addition: construct the table according to
// file size for noticeable speed improvement on small files. Please direct
// questions about this implementation to ames!jaw.
int g_init_bits;
int ClearCode;
int EOFCode;
// output
//
// Output the given code.
// Inputs:
// code: A n_bits-bit integer. If == -1, then EOF. This assumes
// that n_bits =< wordsize - 1.
// Outputs:
// Outputs code to the file.
// Assumptions:
// Chars are 8 bits long.
// Algorithm:
// Maintain a BITS character long buffer (so that 8 codes will
// fit in it exactly). Use the VAX insv instruction to insert each
// code in turn. When the buffer fills up empty it and start over.
int cur_accum = 0;
int cur_bits = 0;
int masks[] =
{
0x0000,
0x0001,
0x0003,
0x0007,
0x000F,
0x001F,
0x003F,
0x007F,
0x00FF,
0x01FF,
0x03FF,
0x07FF,
0x0FFF,
0x1FFF,
0x3FFF,
0x7FFF,
0xFFFF };
// Number of characters so far in this ‘packet‘
int a_count;
// Define the storage for the packet accumulator
byte[] accum = new byte[256];
//----------------------------------------------------------------------------
LZWEncoder(int width, int height, byte[] pixels, int color_depth) {
imgW = width;
imgH = height;
pixAry = pixels;
initCodeSize = Math.max(2, color_depth);
}
// Add a character to the end of the current packet, and if it is 254
// characters, flush the packet to disk.
void char_out(byte c, OutputStream outs) throws IOException {
accum[a_count++] = c;
if (a_count >= 254)
flush_char(outs);
}
// Clear out the hash table
// table clear for block compress
void cl_block(OutputStream outs) throws IOException {
cl_hash(hsize);
free_ent = ClearCode + 2;
clear_flg = true;
output(ClearCode, outs);
}
// reset code table
void cl_hash(int hsize) {
for (int i = 0; i < hsize; ++i)
htab[i] = -1;
}
void compress(int init_bits, OutputStream outs) throws IOException {
int fcode;
int i /* = 0 */;
int c;
int ent;
int disp;
int hsize_reg;
int hshift;
// Set up the globals: g_init_bits - initial number of bits
g_init_bits = init_bits;
// Set up the necessary values
clear_flg = false;
n_bits = g_init_bits;
maxcode = MAXCODE(n_bits);
ClearCode = 1 << (init_bits - 1);
EOFCode = ClearCode + 1;
free_ent = ClearCode + 2;
a_count = 0; // clear packet
ent = nextPixel();
hshift = 0;
for (fcode = hsize; fcode < 65536; fcode *= 2)
++hshift;
hshift = 8 - hshift; // set hash code range bound
hsize_reg = hsize;
cl_hash(hsize_reg); // clear hash table
output(ClearCode, outs);
outer_loop : while ((c = nextPixel()) != EOF) {
fcode = (c << maxbits) + ent;
i = (c << hshift) ^ ent; // xor hashing
if (htab[i] == fcode) {
ent = codetab[i];
continue;
} else if (htab[i] >= 0) // non-empty slot
{
disp = hsize_reg - i; // secondary hash (after G. Knott)
if (i == 0)
disp = 1;
do {
if ((i -= disp) < 0)
i += hsize_reg;
if (htab[i] == fcode) {
ent = codetab[i];
continue outer_loop;
}
} while (htab[i] >= 0);
}
output(ent, outs);
ent = c;
if (free_ent < maxmaxcode) {
codetab[i] = free_ent++; // code -> hashtable
htab[i] = fcode;
} else
cl_block(outs);
}
// Put out the final code.
output(ent, outs);
output(EOFCode, outs);
}
//----------------------------------------------------------------------------
void encode(OutputStream os) throws IOException {
os.write(initCodeSize); // write "initial code size" byte
remaining = imgW * imgH; // reset navigation variables
curPixel = 0;
compress(initCodeSize + 1, os); // compress and write the pixel data
os.write(0); // write block terminator
}
// Flush the packet to disk, and reset the accumulator
void flush_char(OutputStream outs) throws IOException {
if (a_count > 0) {
outs.write(a_count);
outs.write(accum, 0, a_count);
a_count = 0;
}
}
final int MAXCODE(int n_bits) {
return (1 << n_bits) - 1;
}
//----------------------------------------------------------------------------
// Return the next pixel from the image
//----------------------------------------------------------------------------
private int nextPixel() {
if (remaining == 0)
return EOF;
--remaining;
byte pix = pixAry[curPixel++];
return pix & 0xff;
}
void output(int code, OutputStream outs) throws IOException {
cur_accum &= masks[cur_bits];
if (cur_bits > 0)
cur_accum |= (code << cur_bits);
else
cur_accum = code;
cur_bits += n_bits;
while (cur_bits >= 8) {
char_out((byte) (cur_accum & 0xff), outs);
cur_accum >>= 8;
cur_bits -= 8;
}
// If the next entry is going to be too big for the code size,
// then increase it, if possible.
if (free_ent > maxcode || clear_flg) {
if (clear_flg) {
maxcode = MAXCODE(n_bits = g_init_bits);
clear_flg = false;
} else {
++n_bits;
if (n_bits == maxbits)
maxcode = maxmaxcode;
else
maxcode = MAXCODE(n_bits);
}
}
if (code == EOFCode) {
// At EOF, write the rest of the buffer.
while (cur_bits > 0) {
char_out((byte) (cur_accum & 0xff), outs);
cur_accum >>= 8;
cur_bits -= 8;
}
flush_char(outs);
}
}
}
AnimatedGifEncoder.java源码如下：
原文：http://my.oschina.net/u/147181/blog/295607