mirror/JUCE
1
0
Fork 0
mirror of https://github.com/juce-framework/JUCE.git synced 2026-01-10 23:44:24 +00:00
Code Activity

ImageConvolutionKernel: Refactor and fix bugs in single-channel convolution

Browse source
This commit is contained in:
reuk 2024-04-09 13:37:26 +01:00
parent 90f37570f3
commit fae3e2c010
No known key found for this signature in database
GPG key ID: FCB43929F012EE5C
1 changed files with 35 additions and 131 deletions
Download patch file Download diff file Expand all files Collapse all files

166
modules/juce_graphics/images/juce_ImageConvolutionKernel.cpp
View file

@ -42,9 +42,7 @@ ImageConvolutionKernel::ImageConvolutionKernel (int sizeToUse)
clear();
}
ImageConvolutionKernel::~ImageConvolutionKernel()
{
}
ImageConvolutionKernel::~ImageConvolutionKernel() = default;
//==============================================================================
float ImageConvolutionKernel::getKernelValue (const int x, const int y) const noexcept
@ -144,8 +142,10 @@ void ImageConvolutionKernel::applyToImage (Image& destImage,
const Image::BitmapData srcData (sourceImage, Image::BitmapData::readOnly);
if (destData.pixelStride == 4)
const auto applyKernel = [&] (auto stride)
{
constexpr auto pixelStride = stride.value;
for (int y = area.getY(); y < bottom; ++y)
{
uint8* dest = line;
@ -153,10 +153,7 @@ void ImageConvolutionKernel::applyToImage (Image& destImage,
for (int x = area.getX(); x < right; ++x)
{
float c1 = 0;
float c2 = 0;
float c3 = 0;
float c4 = 0;
float sum[pixelStride]{};
for (int yy = 0; yy < size; ++yy)
{
@ -165,140 +162,47 @@ void ImageConvolutionKernel::applyToImage (Image& destImage,
if (sy >= srcData.height)
break;
if (sy >= 0)
if (sy < 0)
continue;
int sx = x - (size >> 1);
const uint8* src = srcData.getPixelPointer (sx, sy);
for (int xx = 0; xx < size; ++xx)
{
int sx = x - (size >> 1);
const uint8* src = srcData.getPixelPointer (sx, sy);
if (sx >= srcData.width)
break;
for (int xx = 0; xx < size; ++xx)
if (sx >= 0)
{
if (sx >= srcData.width)
break;
const auto kernelMult = values[xx + yy * size];
if (sx >= 0)
{
const float kernelMult = values [xx + yy * size];
c1 += kernelMult * *src++;
c2 += kernelMult * *src++;
c3 += kernelMult * *src++;
c4 += kernelMult * *src++;
}
else
{
src += 4;
}
++sx;
for (auto& s : sum)
s += kernelMult * *src++;
}
else
{
src += pixelStride;
}
++sx;
}
}
*dest++ = (uint8) jmin (0xff, roundToInt (c1));
*dest++ = (uint8) jmin (0xff, roundToInt (c2));
*dest++ = (uint8) jmin (0xff, roundToInt (c3));
*dest++ = (uint8) jmin (0xff, roundToInt (c4));
for (const auto& s : sum)
*dest++ = (uint8) jmin (0xff, roundToInt (s));
}
}
}
else if (destData.pixelStride == 3)
};
switch (destData.pixelStride)
{
for (int y = area.getY(); y < bottom; ++y)
{
uint8* dest = line;
line += destData.lineStride;
for (int x = area.getX(); x < right; ++x)
{
float c1 = 0;
float c2 = 0;
float c3 = 0;
for (int yy = 0; yy < size; ++yy)
{
const int sy = y + yy - (size >> 1);
if (sy >= srcData.height)
break;
if (sy >= 0)
{
int sx = x - (size >> 1);
const uint8* src = srcData.getPixelPointer (sx, sy);
for (int xx = 0; xx < size; ++xx)
{
if (sx >= srcData.width)
break;
if (sx >= 0)
{
const float kernelMult = values [xx + yy * size];
c1 += kernelMult * *src++;
c2 += kernelMult * *src++;
c3 += kernelMult * *src++;
}
else
{
src += 3;
}
++sx;
}
}
}
*dest++ = (uint8) roundToInt (c1);
*dest++ = (uint8) roundToInt (c2);
*dest++ = (uint8) roundToInt (c3);
}
}
}
else if (destData.pixelStride == 1)
{
for (int y = area.getY(); y < bottom; ++y)
{
uint8* dest = line;
line += destData.lineStride;
for (int x = area.getX(); x < right; ++x)
{
float c1 = 0;
for (int yy = 0; yy < size; ++yy)
{
const int sy = y + yy - (size >> 1);
if (sy >= srcData.height)
break;
if (sy >= 0)
{
int sx = x - (size >> 1);
const uint8* src = srcData.getPixelPointer (sx, sy);
for (int xx = 0; xx < size; ++xx)
{
if (sx >= srcData.width)
break;
if (sx >= 0)
{
const float kernelMult = values [xx + yy * size];
c1 += kernelMult * *src++;
}
else
{
src += 3;
}
++sx;
}
}
}
*dest++ = (uint8) roundToInt (c1);
}
}
case 4:
return applyKernel (std::integral_constant<size_t, 4>{});
case 3:
return applyKernel (std::integral_constant<size_t, 3>{});
case 1:
return applyKernel (std::integral_constant<size_t, 1>{});
}
}