示例#1
0
void FAST_t(InputArray _img, std::vector<KeyPoint>& keypoints, int threshold, bool nonmax_suppression)
{
    Mat img = _img.getMat();
    const int K = patternSize/2, N = patternSize + K + 1;
#if CV_SSE2
    const int quarterPatternSize = patternSize/4;
    (void)quarterPatternSize;
#endif
    int i, j, k, pixel[25];
    makeOffsets(pixel, (int)img.step, patternSize);

    keypoints.clear();

    threshold = std::min(std::max(threshold, 0), 255);

#if CV_SSE2
    __m128i delta = _mm_set1_epi8(-128), t = _mm_set1_epi8((char)threshold), K16 = _mm_set1_epi8((char)K);
    (void)K16;
    (void)delta;
    (void)t;
#endif
    uchar threshold_tab[512];
    for( i = -255; i <= 255; i++ )
        threshold_tab[i+255] = (uchar)(i < -threshold ? 1 : i > threshold ? 2 : 0);

    AutoBuffer<uchar> _buf((img.cols+16)*3*(sizeof(int) + sizeof(uchar)) + 128);
    uchar* buf[3];
    buf[0] = _buf; buf[1] = buf[0] + img.cols; buf[2] = buf[1] + img.cols;
    int* cpbuf[3];
    cpbuf[0] = (int*)alignPtr(buf[2] + img.cols, sizeof(int)) + 1;
    cpbuf[1] = cpbuf[0] + img.cols + 1;
    cpbuf[2] = cpbuf[1] + img.cols + 1;
    memset(buf[0], 0, img.cols*3);

    for(i = 3; i < img.rows-2; i++)
    {
        const uchar* ptr = img.ptr<uchar>(i) + 3;
        uchar* curr = buf[(i - 3)%3];
        int* cornerpos = cpbuf[(i - 3)%3];
        memset(curr, 0, img.cols);
        int ncorners = 0;

        if( i < img.rows - 3 )
        {
            j = 3;
    #if CV_SSE2
            if( patternSize == 16 )
            {
                for(; j < img.cols - 16 - 3; j += 16, ptr += 16)
                {
                    __m128i m0, m1;
                    __m128i v0 = _mm_loadu_si128((const __m128i*)ptr);
                    __m128i v1 = _mm_xor_si128(_mm_subs_epu8(v0, t), delta);
                    v0 = _mm_xor_si128(_mm_adds_epu8(v0, t), delta);

                    __m128i x0 = _mm_sub_epi8(_mm_loadu_si128((const __m128i*)(ptr + pixel[0])), delta);
                    __m128i x1 = _mm_sub_epi8(_mm_loadu_si128((const __m128i*)(ptr + pixel[quarterPatternSize])), delta);
                    __m128i x2 = _mm_sub_epi8(_mm_loadu_si128((const __m128i*)(ptr + pixel[2*quarterPatternSize])), delta);
                    __m128i x3 = _mm_sub_epi8(_mm_loadu_si128((const __m128i*)(ptr + pixel[3*quarterPatternSize])), delta);
                    m0 = _mm_and_si128(_mm_cmpgt_epi8(x0, v0), _mm_cmpgt_epi8(x1, v0));
                    m1 = _mm_and_si128(_mm_cmpgt_epi8(v1, x0), _mm_cmpgt_epi8(v1, x1));
                    m0 = _mm_or_si128(m0, _mm_and_si128(_mm_cmpgt_epi8(x1, v0), _mm_cmpgt_epi8(x2, v0)));
                    m1 = _mm_or_si128(m1, _mm_and_si128(_mm_cmpgt_epi8(v1, x1), _mm_cmpgt_epi8(v1, x2)));
                    m0 = _mm_or_si128(m0, _mm_and_si128(_mm_cmpgt_epi8(x2, v0), _mm_cmpgt_epi8(x3, v0)));
                    m1 = _mm_or_si128(m1, _mm_and_si128(_mm_cmpgt_epi8(v1, x2), _mm_cmpgt_epi8(v1, x3)));
                    m0 = _mm_or_si128(m0, _mm_and_si128(_mm_cmpgt_epi8(x3, v0), _mm_cmpgt_epi8(x0, v0)));
                    m1 = _mm_or_si128(m1, _mm_and_si128(_mm_cmpgt_epi8(v1, x3), _mm_cmpgt_epi8(v1, x0)));
                    m0 = _mm_or_si128(m0, m1);
                    int mask = _mm_movemask_epi8(m0);
                    if( mask == 0 )
                        continue;
                    if( (mask & 255) == 0 )
                    {
                        j -= 8;
                        ptr -= 8;
                        continue;
                    }

                    __m128i c0 = _mm_setzero_si128(), c1 = c0, max0 = c0, max1 = c0;
                    for( k = 0; k < N; k++ )
                    {
                        __m128i x = _mm_xor_si128(_mm_loadu_si128((const __m128i*)(ptr + pixel[k])), delta);
                        m0 = _mm_cmpgt_epi8(x, v0);
                        m1 = _mm_cmpgt_epi8(v1, x);

                        c0 = _mm_and_si128(_mm_sub_epi8(c0, m0), m0);
                        c1 = _mm_and_si128(_mm_sub_epi8(c1, m1), m1);

                        max0 = _mm_max_epu8(max0, c0);
                        max1 = _mm_max_epu8(max1, c1);
                    }

                    max0 = _mm_max_epu8(max0, max1);
                    int m = _mm_movemask_epi8(_mm_cmpgt_epi8(max0, K16));

                    for( k = 0; m > 0 && k < 16; k++, m >>= 1 )
                        if(m & 1)
                        {
                            cornerpos[ncorners++] = j+k;
                            if(nonmax_suppression)
                                curr[j+k] = (uchar)cornerScore<patternSize>(ptr+k, pixel, threshold);
                        }
                }
            }
    #endif
            for( ; j < img.cols - 3; j++, ptr++ )
            {
                int v = ptr[0];
                const uchar* tab = &threshold_tab[0] - v + 255;
                int d = tab[ptr[pixel[0]]] | tab[ptr[pixel[8]]];

                if( d == 0 )
                    continue;

                d &= tab[ptr[pixel[2]]] | tab[ptr[pixel[10]]];
                d &= tab[ptr[pixel[4]]] | tab[ptr[pixel[12]]];
                d &= tab[ptr[pixel[6]]] | tab[ptr[pixel[14]]];

                if( d == 0 )
                    continue;

                d &= tab[ptr[pixel[1]]] | tab[ptr[pixel[9]]];
                d &= tab[ptr[pixel[3]]] | tab[ptr[pixel[11]]];
                d &= tab[ptr[pixel[5]]] | tab[ptr[pixel[13]]];
                d &= tab[ptr[pixel[7]]] | tab[ptr[pixel[15]]];

                if( d & 1 )
                {
                    int vt = v - threshold, count = 0;

                    for( k = 0; k < N; k++ )
                    {
                        int x = ptr[pixel[k]];
                        if(x < vt)
                        {
                            if( ++count > K )
                            {
                                cornerpos[ncorners++] = j;
                                if(nonmax_suppression)
                                    curr[j] = (uchar)cornerScore<patternSize>(ptr, pixel, threshold);
                                break;
                            }
                        }
                        else
                            count = 0;
                    }
                }

                if( d & 2 )
                {
                    int vt = v + threshold, count = 0;

                    for( k = 0; k < N; k++ )
                    {
                        int x = ptr[pixel[k]];
                        if(x > vt)
                        {
                            if( ++count > K )
                            {
                                cornerpos[ncorners++] = j;
                                if(nonmax_suppression)
                                    curr[j] = (uchar)cornerScore<patternSize>(ptr, pixel, threshold);
                                break;
                            }
                        }
                        else
                            count = 0;
                    }
                }
            }
        }

        cornerpos[-1] = ncorners;

        if( i == 3 )
            continue;

        const uchar* prev = buf[(i - 4 + 3)%3];
        const uchar* pprev = buf[(i - 5 + 3)%3];
        cornerpos = cpbuf[(i - 4 + 3)%3];
        ncorners = cornerpos[-1];

        for( k = 0; k < ncorners; k++ )
        {
            j = cornerpos[k];
            int score = prev[j];
            if( !nonmax_suppression ||
               (score > prev[j+1] && score > prev[j-1] &&
                score > pprev[j-1] && score > pprev[j] && score > pprev[j+1] &&
                score > curr[j-1] && score > curr[j] && score > curr[j+1]) )
            {
                keypoints.push_back(KeyPoint((float)j, (float)(i-1), 7.f, -1, (float)score));
            }
        }
    }
示例#2
0
void FAST_t(InputArray _img, std::vector<KeyPoint>& keypoints, int threshold, bool nonmax_suppression)
{
    Mat img = _img.getMat();
    const int K = patternSize/2, N = patternSize + K + 1;
    int i, j, k, pixel[25];
    makeOffsets(pixel, (int)img.step, patternSize);

#if CV_SIMD128
    const int quarterPatternSize = patternSize/4;
    v_uint8x16 delta = v_setall_u8(0x80), t = v_setall_u8((char)threshold), K16 = v_setall_u8((char)K);
    bool hasSimd = hasSIMD128();
#if CV_TRY_AVX2
    Ptr<opt_AVX2::FAST_t_patternSize16_AVX2> fast_t_impl_avx2;
    if(CV_CPU_HAS_SUPPORT_AVX2)
        fast_t_impl_avx2 = opt_AVX2::FAST_t_patternSize16_AVX2::getImpl(img.cols, threshold, nonmax_suppression, pixel);
#endif

#endif

    keypoints.clear();

    threshold = std::min(std::max(threshold, 0), 255);

    uchar threshold_tab[512];
    for( i = -255; i <= 255; i++ )
        threshold_tab[i+255] = (uchar)(i < -threshold ? 1 : i > threshold ? 2 : 0);

    AutoBuffer<uchar> _buf((img.cols+16)*3*(sizeof(int) + sizeof(uchar)) + 128);
    uchar* buf[3];
    buf[0] = _buf.data(); buf[1] = buf[0] + img.cols; buf[2] = buf[1] + img.cols;
    int* cpbuf[3];
    cpbuf[0] = (int*)alignPtr(buf[2] + img.cols, sizeof(int)) + 1;
    cpbuf[1] = cpbuf[0] + img.cols + 1;
    cpbuf[2] = cpbuf[1] + img.cols + 1;
    memset(buf[0], 0, img.cols*3);

    for(i = 3; i < img.rows-2; i++)
    {
        const uchar* ptr = img.ptr<uchar>(i) + 3;
        uchar* curr = buf[(i - 3)%3];
        int* cornerpos = cpbuf[(i - 3)%3];
        memset(curr, 0, img.cols);
        int ncorners = 0;

        if( i < img.rows - 3 )
        {
            j = 3;
#if CV_SIMD128
            if( hasSimd )
            {
                if( patternSize == 16 )
                {
#if CV_TRY_AVX2
                    if (fast_t_impl_avx2)
                        fast_t_impl_avx2->process(j, ptr, curr, cornerpos, ncorners);
#endif
                    //vz if (j <= (img.cols - 27)) //it doesn't make sense using vectors for less than 8 elements
                    {
                        for (; j < img.cols - 16 - 3; j += 16, ptr += 16)
                        {
                            v_uint8x16 v = v_load(ptr);
                            v_int8x16 v0 = v_reinterpret_as_s8((v + t) ^ delta);
                            v_int8x16 v1 = v_reinterpret_as_s8((v - t) ^ delta);

                            v_int8x16 x0 = v_reinterpret_as_s8(v_sub_wrap(v_load(ptr + pixel[0]), delta));
                            v_int8x16 x1 = v_reinterpret_as_s8(v_sub_wrap(v_load(ptr + pixel[quarterPatternSize]), delta));
                            v_int8x16 x2 = v_reinterpret_as_s8(v_sub_wrap(v_load(ptr + pixel[2*quarterPatternSize]), delta));
                            v_int8x16 x3 = v_reinterpret_as_s8(v_sub_wrap(v_load(ptr + pixel[3*quarterPatternSize]), delta));

                            v_int8x16 m0, m1;
                            m0 = (v0 < x0) & (v0 < x1);
                            m1 = (x0 < v1) & (x1 < v1);
                            m0 = m0 | ((v0 < x1) & (v0 < x2));
                            m1 = m1 | ((x1 < v1) & (x2 < v1));
                            m0 = m0 | ((v0 < x2) & (v0 < x3));
                            m1 = m1 | ((x2 < v1) & (x3 < v1));
                            m0 = m0 | ((v0 < x3) & (v0 < x0));
                            m1 = m1 | ((x3 < v1) & (x0 < v1));
                            m0 = m0 | m1;

                            int mask = v_signmask(m0);
                            if( mask == 0 )
                                continue;
                            if( (mask & 255) == 0 )
                            {
                                j -= 8;
                                ptr -= 8;
                                continue;
                            }

                            v_int8x16 c0 = v_setzero_s8();
                            v_int8x16 c1 = v_setzero_s8();
                            v_uint8x16 max0 = v_setzero_u8();
                            v_uint8x16 max1 = v_setzero_u8();
                            for( k = 0; k < N; k++ )
                            {
                                v_int8x16 x = v_reinterpret_as_s8(v_load((ptr + pixel[k])) ^ delta);
                                m0 = v0 < x;
                                m1 = x < v1;

                                c0 = v_sub_wrap(c0, m0) & m0;
                                c1 = v_sub_wrap(c1, m1) & m1;

                                max0 = v_max(max0, v_reinterpret_as_u8(c0));
                                max1 = v_max(max1, v_reinterpret_as_u8(c1));
                            }

                            max0 = v_max(max0, max1);
                            int m = v_signmask(K16 < max0);

                            for( k = 0; m > 0 && k < 16; k++, m >>= 1 )
                            {
                                if(m & 1)
                                {
                                    cornerpos[ncorners++] = j+k;
                                    if(nonmax_suppression)
                                        curr[j+k] = (uchar)cornerScore<patternSize>(ptr+k, pixel, threshold);
                                }
                            }
                        }
                    }
                }
            }
#endif
            for( ; j < img.cols - 3; j++, ptr++ )
            {
                int v = ptr[0];
                const uchar* tab = &threshold_tab[0] - v + 255;
                int d = tab[ptr[pixel[0]]] | tab[ptr[pixel[8]]];

                if( d == 0 )
                    continue;

                d &= tab[ptr[pixel[2]]] | tab[ptr[pixel[10]]];
                d &= tab[ptr[pixel[4]]] | tab[ptr[pixel[12]]];
                d &= tab[ptr[pixel[6]]] | tab[ptr[pixel[14]]];

                if( d == 0 )
                    continue;

                d &= tab[ptr[pixel[1]]] | tab[ptr[pixel[9]]];
                d &= tab[ptr[pixel[3]]] | tab[ptr[pixel[11]]];
                d &= tab[ptr[pixel[5]]] | tab[ptr[pixel[13]]];
                d &= tab[ptr[pixel[7]]] | tab[ptr[pixel[15]]];

                if( d & 1 )
                {
                    int vt = v - threshold, count = 0;

                    for( k = 0; k < N; k++ )
                    {
                        int x = ptr[pixel[k]];
                        if(x < vt)
                        {
                            if( ++count > K )
                            {
                                cornerpos[ncorners++] = j;
                                if(nonmax_suppression)
                                    curr[j] = (uchar)cornerScore<patternSize>(ptr, pixel, threshold);
                                break;
                            }
                        }
                        else
                            count = 0;
                    }
                }

                if( d & 2 )
                {
                    int vt = v + threshold, count = 0;

                    for( k = 0; k < N; k++ )
                    {
                        int x = ptr[pixel[k]];
                        if(x > vt)
                        {
                            if( ++count > K )
                            {
                                cornerpos[ncorners++] = j;
                                if(nonmax_suppression)
                                    curr[j] = (uchar)cornerScore<patternSize>(ptr, pixel, threshold);
                                break;
                            }
                        }
                        else
                            count = 0;
                    }
                }
            }
        }

        cornerpos[-1] = ncorners;

        if( i == 3 )
            continue;

        const uchar* prev = buf[(i - 4 + 3)%3];
        const uchar* pprev = buf[(i - 5 + 3)%3];
        cornerpos = cpbuf[(i - 4 + 3)%3];
        ncorners = cornerpos[-1];

        for( k = 0; k < ncorners; k++ )
        {
            j = cornerpos[k];
            int score = prev[j];
            if( !nonmax_suppression ||
               (score > prev[j+1] && score > prev[j-1] &&
                score > pprev[j-1] && score > pprev[j] && score > pprev[j+1] &&
                score > curr[j-1] && score > curr[j] && score > curr[j+1]) )
            {
                keypoints.push_back(KeyPoint((float)j, (float)(i-1), 7.f, -1, (float)score));
            }
        }
    }
示例#3
0
MetadataFromBlr::MetadataFromBlr(unsigned aBlrLength, const unsigned char* aBlr, unsigned aLength)
{
	if (aBlrLength == 0)
		return;

	BlrReader rdr(aBlr, aBlrLength);

	const UCHAR byte = rdr.getByte();
	if (byte != blr_version4 && byte != blr_version5)
	{
		(Arg::Gds(isc_dsql_error) <<
		 Arg::Gds(isc_sqlerr) << Arg::Num(-804) <<
		 Arg::Gds(isc_wroblrver2) << Arg::Num(blr_version4) << Arg::Num(blr_version5) << Arg::Num(byte)
		).raise();
	}

	if (rdr.getByte() != blr_begin || rdr.getByte() != blr_message)
	{
		(Arg::Gds(isc_sqlerr) << Arg::Num(-804) <<
		 Arg::Gds(isc_dsql_sqlda_err)
#ifdef DEV_BUILD
		 << Arg::Gds(isc_random) << "Missing blr_begin / blr_message"
#endif
		).raise();
	}

	rdr.getByte();						// skip the message number
	unsigned count = rdr.getWord();
	fb_assert(!(count & 1));
	count /= 2;

	unsigned offset = 0;
	items.grow(count);

	for (unsigned index = 0; index < count; index++)
	{
		Item* item = &items[index];
		item->scale = 0;
		item->subType = 0;

		switch (rdr.getByte())
		{
		case blr_text:
			item->type = SQL_TEXT;
			item->charSet = CS_dynamic;
			item->length = rdr.getWord();
			break;

		case blr_varying:
			item->type = SQL_VARYING;
			item->charSet = CS_dynamic;
			item->length = rdr.getWord();
			break;

		case blr_text2:
			item->type = SQL_TEXT;
			item->charSet = rdr.getWord();
			item->length = rdr.getWord();
			break;

		case blr_varying2:
			item->type = SQL_VARYING;
			item->charSet = rdr.getWord();
			item->length = rdr.getWord();
			break;

		case blr_short:
			item->type = SQL_SHORT;
			item->length = sizeof(SSHORT);
			item->scale = rdr.getByte();
			break;

		case blr_long:
			item->type = SQL_LONG;
			item->length = sizeof(SLONG);
			item->scale = rdr.getByte();
			break;

		case blr_int64:
			item->type = SQL_INT64;
			item->length = sizeof(SINT64);
			item->scale = rdr.getByte();
			break;

		case blr_quad:
			item->type = SQL_QUAD;
			item->length = sizeof(SLONG) * 2;
			item->scale = rdr.getByte();
			break;

		case blr_float:
			item->type = SQL_FLOAT;
			item->length = sizeof(float);
			break;

		case blr_double:
		case blr_d_float:
			item->type = SQL_DOUBLE;
			item->length = sizeof(double);
			break;

		case blr_timestamp:
			item->type = SQL_TIMESTAMP;
			item->length = sizeof(SLONG) * 2;
			break;

		case blr_sql_date:
			item->type = SQL_TYPE_DATE;
			item->length = sizeof(SLONG);
			break;

		case blr_sql_time:
			item->type = SQL_TYPE_TIME;
			item->length = sizeof(SLONG);
			break;

		case blr_blob2:
			item->type = SQL_BLOB;
			item->length = sizeof(ISC_QUAD);
			item->subType = rdr.getWord();
			item->charSet = rdr.getWord();
			break;

		case blr_bool:
			item->type = SQL_BOOLEAN;
			item->length = sizeof(UCHAR);
			break;

		default:
			(Arg::Gds(isc_sqlerr) << Arg::Num(-804) <<
			 Arg::Gds(isc_dsql_sqlda_err)
#ifdef DEV_BUILD
			 << Arg::Gds(isc_random) << "Wrong BLR type"
#endif
			).raise();
		}

		if (rdr.getByte() != blr_short || rdr.getByte() != 0)
		{
			(Arg::Gds(isc_sqlerr) << Arg::Num(-804) <<
			 Arg::Gds(isc_dsql_sqlda_err)
#ifdef DEV_BUILD
			 << Arg::Gds(isc_random) << "Wrong BLR type for NULL indicator"
#endif
			).raise();
		}

		item->finished = true;
	}

	makeOffsets();

	if (rdr.getByte() != (UCHAR) blr_end || length != aLength)
	{
		(Arg::Gds(isc_sqlerr) << Arg::Num(-804) <<
		 Arg::Gds(isc_dsql_sqlda_err)
#ifdef DEV_BUILD
		 << Arg::Gds(isc_random) << (length != aLength ? "Invalid message length" : "Missing blr_end")
#endif
		).raise();
	}
}