Small JPEG optimizations.

This commit is contained in:
iabdalkader 2016-02-12 07:08:22 +02:00
parent 3acb5903a2
commit 9685cbb727

View File

@ -24,10 +24,11 @@ typedef struct {
int idx;
int length;
uint8_t *buf;
int bitc, bitb;
} jpeg_buf_t;
// Quantization tables
static float fdtbl_Y[64], fdtbl_UV[64];
static int fdtbl_Y[64], fdtbl_UV[64];
static uint8_t YTable[64], UVTable[64];
// RGB565 to YUV table
@ -191,38 +192,34 @@ static void jpeg_put_bytes(jpeg_buf_t *jpeg_buf, const void *data, int size)
jpeg_buf->idx += size;
}
static void jpeg_writeBits(jpeg_buf_t *jpeg_buf, int *bitBuf, int *bitCnt, const uint16_t *bs)
static void jpeg_writeBits(jpeg_buf_t *jpeg_buf, const uint16_t *bs)
{
int bitc = *bitCnt;
int bitb = *bitBuf;
bitc += bs[1];
bitb |= bs[0] << (24 - bitc);
jpeg_buf->bitc += bs[1];
jpeg_buf->bitb |= bs[0] << (24 - jpeg_buf->bitc);
while (bitc > 7) {
uint8_t c = (bitb >> 16) & 255;
while (jpeg_buf->bitc > 7) {
uint8_t c = (jpeg_buf->bitb >> 16) & 255;
jpeg_put_char(jpeg_buf, c);
if(c == 255) {
jpeg_put_char(jpeg_buf, 0);
}
bitb <<= 8;
bitc -= 8;
jpeg_buf->bitb <<= 8;
jpeg_buf->bitc -= 8;
}
*bitCnt = bitc;
*bitBuf = bitb;
}
//Huffman-encoded magnitude value
static void jpeg_calcBits(int val, uint16_t bits[2]) {
int t1 = val < 0 ? -val : val;
val = val < 0 ? val-1 : val;
bits[1] = 1;
while(t1 >>= 1) {
++bits[1];
int t1=val;
if (val<0) {
t1 = -val;
val = val-1;
}
bits[1] = 32-__CLZ(t1);
bits[0] = val & ((1<<bits[1])-1);
}
static int jpeg_processDU(jpeg_buf_t *jpeg_buf, int *bitBuf, int *bitCnt, int *CDU,
float *fdtbl, int DC, const uint16_t (*HTDC)[2], const uint16_t (*HTAC)[2])
static int jpeg_processDU(jpeg_buf_t *jpeg_buf, int16_t *CDU, int *fdtbl, int DC, const uint16_t (*HTDC)[2], const uint16_t (*HTAC)[2])
{
const uint16_t EOB[2] = { HTAC[0x00][0], HTAC[0x00][1] };
const uint16_t M16zeroes[2] = { HTAC[0xF0][0], HTAC[0xF0][1] };
@ -232,7 +229,7 @@ static int jpeg_processDU(jpeg_buf_t *jpeg_buf, int *bitBuf, int *bitCnt, int *C
// BinDCT
// DCT rows
for (int i=8, *p=CDU; i>0; i--, p+=8) {
for (int16_t i=8, *p=CDU; i>0; i--, p+=8) {
t0 = p[0] + p[7];
t1 = p[1] + p[6];
t2 = p[2] + p[5];
@ -275,7 +272,7 @@ static int jpeg_processDU(jpeg_buf_t *jpeg_buf, int *bitBuf, int *bitCnt, int *C
}
// DCT columns
for (int i=8, *p=CDU; i>0; i--, p++) {
for (int16_t i=8, *p=CDU; i>0; i--, p++) {
t0 = p[0] + p[56];
t1 = p[8] + p[48];
t2 = p[16] + p[40];
@ -317,30 +314,32 @@ static int jpeg_processDU(jpeg_buf_t *jpeg_buf, int *bitBuf, int *bitCnt, int *C
p[56] = z11 - z4;
}
// first non-zero element in reverse order
int end0pos = 0;
// Quantize/descale/zigzag the coefficients
int DU[64];
for(int i=0; i<64; ++i) {
DU[s_jpeg_ZigZag[i]] = (int)(CDU[i]*fdtbl[i]);
DU[s_jpeg_ZigZag[i]] = (CDU[i]*fdtbl[i])/4096;
if (s_jpeg_ZigZag[i] > end0pos && DU[s_jpeg_ZigZag[i]]) {
end0pos = s_jpeg_ZigZag[i];
}
}
// Encode DC
int diff = DU[0] - DC;
if (diff == 0) {
jpeg_writeBits(jpeg_buf, bitBuf, bitCnt, HTDC[0]);
jpeg_writeBits(jpeg_buf, HTDC[0]);
} else {
uint16_t bits[2];
jpeg_calcBits(diff, bits);
jpeg_writeBits(jpeg_buf, bitBuf, bitCnt, HTDC[bits[1]]);
jpeg_writeBits(jpeg_buf, bitBuf, bitCnt, bits);
jpeg_writeBits(jpeg_buf, HTDC[bits[1]]);
jpeg_writeBits(jpeg_buf, bits);
}
// Encode ACs
int end0pos = 63;
for(; (end0pos>0)&&(DU[end0pos]==0); --end0pos) {
}
// end0pos = first element in reverse order !=0
if(end0pos == 0) {
jpeg_writeBits(jpeg_buf, bitBuf, bitCnt, EOB);
jpeg_writeBits(jpeg_buf, EOB);
return DU[0];
}
for(int i = 1; i <= end0pos; ++i) {
@ -351,16 +350,16 @@ static int jpeg_processDU(jpeg_buf_t *jpeg_buf, int *bitBuf, int *bitCnt, int *C
if ( nrzeroes >= 16 ) {
int lng = nrzeroes>>4;
for (int nrmarker=1; nrmarker <= lng; ++nrmarker)
jpeg_writeBits(jpeg_buf, bitBuf, bitCnt, M16zeroes);
jpeg_writeBits(jpeg_buf, M16zeroes);
nrzeroes &= 15;
}
uint16_t bits[2];
jpeg_calcBits(DU[i], bits);
jpeg_writeBits(jpeg_buf, bitBuf, bitCnt, HTAC[(nrzeroes<<4)+bits[1]]);
jpeg_writeBits(jpeg_buf, bitBuf, bitCnt, bits);
jpeg_writeBits(jpeg_buf, HTAC[(nrzeroes<<4)+bits[1]]);
jpeg_writeBits(jpeg_buf, bits);
}
if(end0pos != 63) {
jpeg_writeBits(jpeg_buf, bitBuf, bitCnt, EOB);
jpeg_writeBits(jpeg_buf, EOB);
}
return DU[0];
}
@ -383,8 +382,8 @@ void jpeg_init(int quality)
for(int r = 0, k = 0; r < 8; ++r) {
for(int c = 0; c < 8; ++c, ++k) {
fdtbl_Y[k] = 1 / (YTable [s_jpeg_ZigZag[k]] * aasf[r] * aasf[c] * 8.0f);
fdtbl_UV[k] = 1 / (UVTable[s_jpeg_ZigZag[k]] * aasf[r] * aasf[c] * 8.0f);
fdtbl_Y[k] = (int)((1/(YTable [s_jpeg_ZigZag[k]] * aasf[r] * aasf[c] * 8.0f))*4096);
fdtbl_UV[k] = (int)((1/(UVTable[s_jpeg_ZigZag[k]] * aasf[r] * aasf[c] * 8.0f))*4096);
}
}
}
@ -425,15 +424,16 @@ void jpeg_write_headers(jpeg_buf_t *jpeg_buf, int width, int height)
void jpeg_compress(image_t *src, image_t *dst, int quality)
{
int bitBuf=0, bitCnt=0;
int DCY=0, DCU=0, DCV=0;
int YDU[64], UDU[64], VDU[64];
int16_t YDU[64], UDU[64], VDU[64];
// JPEG buffer
jpeg_buf_t jpeg_buf = {
.idx =0,
.buf = dst->pixels,
.length = dst->bpp,
.bitc = 0,
.bitb = 0,
};
// Initialize quantization tables
@ -461,14 +461,14 @@ void jpeg_compress(image_t *src, image_t *dst, int quality)
YDU[pos + 7] = pixels[ofs + 7] - 128;
}
DCY = jpeg_processDU(&jpeg_buf, &bitBuf, &bitCnt, YDU, fdtbl_Y, DCY, YDC_HT, YAC_HT);
DCY = jpeg_processDU(&jpeg_buf, YDU, fdtbl_Y, DCY, YDC_HT, YAC_HT);
// Skip UV blocks
jpeg_writeBits(&jpeg_buf, &bitBuf, &bitCnt, UVDC_HT[0]);
jpeg_writeBits(&jpeg_buf, &bitBuf, &bitCnt, EOB);
jpeg_writeBits(&jpeg_buf, UVDC_HT[0]);
jpeg_writeBits(&jpeg_buf, EOB);
jpeg_writeBits(&jpeg_buf, &bitBuf, &bitCnt, UVDC_HT[0]);
jpeg_writeBits(&jpeg_buf, &bitBuf, &bitCnt, EOB);
jpeg_writeBits(&jpeg_buf, UVDC_HT[0]);
jpeg_writeBits(&jpeg_buf, EOB);
}
}
} else if (src->bpp == 2) {// TODO assuming RGB565
@ -510,16 +510,16 @@ void jpeg_compress(image_t *src, image_t *dst, int quality)
VDU[pos + 7] = yuv_table[pixels[ofs + 7] * 3 + 2];
}
DCY = jpeg_processDU(&jpeg_buf, &bitBuf, &bitCnt, YDU, fdtbl_Y, DCY, YDC_HT, YAC_HT);
DCU = jpeg_processDU(&jpeg_buf, &bitBuf, &bitCnt, UDU, fdtbl_UV, DCU, UVDC_HT, UVAC_HT);
DCV = jpeg_processDU(&jpeg_buf, &bitBuf, &bitCnt, VDU, fdtbl_UV, DCV, UVDC_HT, UVAC_HT);
DCY = jpeg_processDU(&jpeg_buf, YDU, fdtbl_Y, DCY, YDC_HT, YAC_HT);
DCU = jpeg_processDU(&jpeg_buf, UDU, fdtbl_UV, DCU, UVDC_HT, UVAC_HT);
DCV = jpeg_processDU(&jpeg_buf, VDU, fdtbl_UV, DCV, UVDC_HT, UVAC_HT);
}
}
}
// Do the bit alignment of the EOI marker
static const uint16_t fillBits[] = {0x7F, 7};
jpeg_writeBits(&jpeg_buf, &bitBuf, &bitCnt, fillBits);
jpeg_writeBits(&jpeg_buf, fillBits);
// EOI
jpeg_put_char(&jpeg_buf, 0xFF);