Develop Reference

Concatenate a bitmap (rgb) with a TIFF (cmyk) without converting cmyk to rgb

I'm developing an application to concatenate a bitmap image in RGB with a TIFF in CMYK.
I've tried with System.Drawing and System.Windows.Media namespaces.
The problem is both the libraries try to convert my TIFF image into RGB before merging, which causes a loss in image quality.
As far as I understand, the reason they always convert images into RGB before processing because the two libraries do that with a rendering intent.
I don't need to render anything, just merge the two photos and save to disk, that's all.
What should I do to achieve my goal? Clearly, I don't want to lose the quality of the TIFF so I think it's best to not do any conversion, just keep it raw and merge. Anyway, that's just a guess, other option could be considered as well. Could anybody shed some light on my case please?
See a comparison of the tiff image before and after converted from cmyk to rgb below.

I’m not aware of any capacity in the TIFF format to have two different color spaces at the same time. Since you are dealing in CMYK, I assume that is the one you want to preserve.
If so, the steps to do so would be:
Load CMYK image A (using BitmapDecoder)
Load RGB image B (using BitmapDecoder)
Convert image B to CMYK with the desired color profile (using FormatConvertedBitmap)
If required, ensure the pixel format for image B matches A (using FormatConvertedBitmap)
Composite the two in memory as a byte array (using CopyPixels, then memory manipulation, then new bitmap from the memory)
Save the composite to a new CMYK TIFF file (using TiffBitmapEncoder)
That should be possible with WIC (System.Media).
An example doing so (github) could be written as:
BitmapFrame LoadTiff(string filename)
{
using (var rs = File.OpenRead(filename))
{
return BitmapDecoder.Create(rs, BitmapCreateOptions.PreservePixelFormat, BitmapCacheOption.OnLoad).Frames[0];
}
}
// Load, validate A
var imageA = LoadTiff("CMYK.tif");
if (imageA.Format != PixelFormats.Cmyk32)
{
throw new InvalidOperationException("imageA is not CMYK");
}
// Load, validate, convert B
var imageB = LoadTiff("RGB.tif");
if (imageB.PixelHeight != imageA.PixelHeight)
{
throw new InvalidOperationException("Image B is not the same height as image A");
}
var imageBCmyk = new FormatConvertedBitmap(imageB, imageA.Format, null, 0d);
// Merge
int width = imageA.PixelWidth + imageB.PixelWidth,
height = imageA.PixelHeight,
bytesPerPixel = imageA.Format.BitsPerPixel / 8,
stride = width * bytesPerPixel;
var buffer = new byte[stride * height];
imageA.CopyPixels(buffer, stride, 0);
imageBCmyk.CopyPixels(buffer, stride, imageA.PixelWidth * bytesPerPixel);
var result = BitmapSource.Create(width, height, imageA.DpiX, imageA.DpiY, imageA.Format, null, buffer, stride);
// save to new file
using (var ws = File.Create("out.tif"))
{
var tiffEncoder = new TiffBitmapEncoder();
tiffEncoder.Frames.Add(BitmapFrame.Create(result));
tiffEncoder.Save(ws);
}
Which maintains color accuracy of the CMYK image, and converts the RGB using the system color profile. This can be verified in Photoshop which shows that the each letter, and rich black, have maintained their original values. (note that imgur does convert to png with dubious color handling - check github for originals.)
Image A (CMYK):
Image B (RGB):
Result (CMYK):
To have the two images overlayed, one image would have to have some notion of transparency. A mask would be one example thereof, where you pick a particular color value to mean "transparent". The downside of a mask is that masks do not play well with aliased source images. For that, you would want to do an alpha channel - but blending across color spaces would be challenging. (Github)
// Load, validate A
var imageA = LoadTiff("CMYK.tif");
if (imageA.Format != PixelFormats.Cmyk32)
{
throw new InvalidOperationException("imageA is not CMYK");
}
// Load, validate, convert B
var imageB = LoadTiff("RGBOverlay.tif");
if (imageB.PixelHeight != imageA.PixelHeight
|| imageB.PixelWidth != imageA.PixelWidth)
{
throw new InvalidOperationException("Image B is not the same size as image A");
}
var imageBBGRA = new FormatConvertedBitmap(imageB, PixelFormats.Bgra32, null, 0d);
var imageBCmyk = new FormatConvertedBitmap(imageB, imageA.Format, null, 0d);
// Merge
int width = imageA.PixelWidth, height = imageA.PixelHeight;
var stride = width * (imageA.Format.BitsPerPixel / 8);
var bufferA = new uint[width * height];
var bufferB = new uint[width * height];
var maskBuffer = new uint[width * height];
imageA.CopyPixels(bufferA, stride, 0);
imageBBGRA.CopyPixels(maskBuffer, stride, 0);
imageBCmyk.CopyPixels(bufferB, stride, 0);
for (int i = 0; i < bufferA.Length; i++)
{
// set pixel in bufferA to the value from bufferB if mask is not white
if (maskBuffer[i] != 0xffffffff)
{
bufferA[i] = bufferB[i];
}
}
var result = BitmapSource.Create(width, height, imageA.DpiX, imageA.DpiY, imageA.Format, null, bufferA, stride);
// save to new file
using (var ws = File.Create("out_overlay.tif"))
{
var tiffEncoder = new TiffBitmapEncoder();
tiffEncoder.Frames.Add(BitmapFrame.Create(result));
tiffEncoder.Save(ws);
}
Example image B:
Example output:

Why is AForge.net giving a different output in case of FFT Auto-correlation?

See this related question.
I want to obtain the same outcome using AForge.net framework. The output should match the following:
The output seems to be not coming as expected:
Why is the output different in AForge.net?
.
Source Code
public partial class Form1 : Form
{
public Form1()
{
InitializeComponent();
Bitmap image = (Bitmap)Bitmap.FromFile(#"StandardImage\\lena.png");
Bitmap conv = new Bitmap(image.Width, image.Height, image.PixelFormat);
ComplexImage cImage = ComplexImage.FromBitmap(image);
cImage.ForwardFourierTransform();
ComplexImage cKernel = ComplexImage.FromBitmap(image);
cImage.ForwardFourierTransform();
ComplexImage convOut = ComplexImage.FromBitmap(conv);
convOut.ForwardFourierTransform();
for (int y = 0; y < cImage.Height; y++)
{
for (int x = 0; x < cImage.Width; x++)
{
convOut.Data[x, y] = cImage.Data[x, y] * cKernel.Data[x, y];
}
}
convOut.BackwardFourierTransform();
Bitmap bbbb = convOut.ToBitmap();
pictureBox1.Image = bbbb;
}
}

The main problem is
ComplexImage cKernel = ComplexImage.FromBitmap(image);
//cImage.ForwardFourierTransform(); //<--- This line should be FFT of cKernel
cKernel.ForwardFourierTransform();
This would solve the problem you mentioned in the resulting image, but if you want to get an image similar to the bottom right image you need to do some normalization to increase the intensity of pixels.
update:
The bottom right image is actually a Fourier image so I think we should remove the BFF.
//convOut.BackwardFourierTransform();

It is seems like you are not using a gaussian kernel with Aforge,
Anyway, the library has a method for convolution with gaussian:
int w=4,h=11;
GaussianBlur filter = new GaussianBlur( w, h );
// apply the filter
filter.ApplyInPlace( image );
Try it and the output should be the same as the others.

Looking for example for plotting Volumetric Slice in Ilnumerics

is there any example for plotting volumetric slice in Ilnumerics use community version. This is an example I got from matlab website:
Volumetric slice image example of matlab
I have array X, Y, Z as posistions and V (velocity) as value for color plotting. All I have done is use Ilpoints to plot that V in position X, Y, Z not , a surfaces. Here are My Code and the result,
ILArray<float> plotXY = ILMath.zeros<float>(3, XcoordinateXY.Length);
plotXY["0;:"] = ILMath.tosingle(SurfaceXY[":;:;1"]);
plotXY["1;:"] = ILMath.tosingle(SurfaceXY[":;:;2"]);
plotXY["2;:"] = ILMath.tosingle(SurfaceXY[":;:;3"]);
ILArray<float> ColorMap = ILMath.tosingle(SurfaceXY[":;:;0"]);
var ilsurfaceplotXY = new ILPoints()
{
/*Wireframe = { Color = Color.FromArgb(50, Color.LightGray) },
Colormap = new ILColormap(dataXY),
Children = { new ILColorbar() }*/
Positions = plotXY,
Colors = cm.Map(ColorMap).T,
Color = null
};
Here are code for displaying:
var scene = new ILScene();
scene.Add(
new ILPlotCube
{
TwoDMode = false,
Axes =
{
XAxis =
{
Label = { Text = "UTM X (Km)" },
GridMajor =
{
DashStyle = DashStyle.Dashed,
Color = Color.DarkGray,
Width = 1
}
},
YAxis =
{
Label = { Text = "UTM Y (Km)" },
GridMajor =
{
DashStyle = DashStyle.Dashed,
Color = Color.DarkGray,
Width = 1
}
},
ZAxis =
{
Label = { Text = "DEPTH (Km)" },
GridMajor =
{
DashStyle = DashStyle.Dashed,
Color = Color.DarkGray,
Width = 1
}
}
},
Children = { ilsurfaceplotXY, ilsurfaceplotXZ, ilsurfaceplotYZ },
}
);
this.ilPanel1.Scene = scene;
this.ilPanel1.Scene.Configure();
this.ilPanel1.Refresh();
And here is an image result.
Result Image
I'm sorry the image is in the link.

Regarding the visualization this can be done with regular surfaces, imagesc plots, or the new fast surface in the Drawing2 toolbox. They all allow to provide X,Y, and Z values as well as a color for each grid point or tile.
Regarding the computation of the points: it seems that you just pick points from the available set. It would be much better to interpolate between these points. The Interpolation Toolbox provides functions for the interpolation of gridded and scattered data. (In your case the data seem to be gridded ?). This allows to have slices in arbitrary orientation / angles. The interpolation toolbox interpolates the positions of the slice grid points as well as the values for the colors.
From an online example:
The setup of the horizontal slices is done as follows:
ILArray<float> C;
for (int i = 0; i < m_nrSlices; i += m_nrSlices / 4) {
C = m_V[":",":", i];
pc1.Add(new ILSurface(grid + i, C, colormap: Colormaps.Bone)
{
Wireframe = { Visible = false },
});
}
Here, m_V is your 3D dataset, handled as 3D array. pc is the plot cube. The surfaces are simply added to the plot cube. The points of the red interpolated area are dynamically computed as the user moves the red balls:
// Points on the cutting area are considered scattered points, because the area is not (necessarily) plain. However, V
// is a grid. interp3s interpolates the scattered points very efficiently.
// Note how the shape of the coordinate arrays Xn, Yn and Zn is not important. interp3s takes their elements in sequential order.
// The output is a vector of interpolated values. (We gonna reshape it below.)
ILArray < float> Z = Interpolation.interp3s(m_V, m_x, m_x, m_x, m_Xn, m_Yn, Zn, method: InterpolationMethod.cubic);
// let's plot! We get a reference to the fast surface
var fsurf = ilPanel1.Scene.First<ILFastSurface>("dynslice");
if (fsurf != null) {
// first time setup only: provide the full coordinates of X and V. Here it is sufficient to provide grid vectors.
if (fsurf.Cols == 0) {
fsurf.Update(X: m_xn * res, Y: m_xn * res, Z: Zn * res, C: ILMath.reshape(Z, Zn.S), colormap: Colormaps.Hot);
} else {
// the grid was configured already and did not change. we save some recomputing by ommiting the X and Y coordinates, prevent from reshaping buffers.
fsurf.Update(Z: Zn * res, C: ILMath.reshape(Z, Zn.S), colormap: Colormaps.Hot);
}
}
fsurf.Configure();
ilPanel1.Refresh();
To go into the details is out of scope for SO. You can download the example and run it on your machine. You will need a recent version of ILNumerics though.
EDIT: Axis aligned slices as in the plot you provided are only a subdomain, of course. Generating them works in the very same way:

Is it possible to find the edge of a "spotty" region in emgucv?

I have an image that looks like this:
and I want to find the edges of the dark part so like this (the red lines are what I am looking for):
I have tried a few approaches and none have worked so I am hoping there is an emgu guru out there willing to help me...
Approach 1
Convert the image to grayscale
Remove noise and invert
Remove anything that is not really bright
Get the canny and the polygons
Code for this (I know that I should be disposing of things properly but I am keeping the code short):
var orig = new Image<Bgr, byte>(inFile);
var contours = orig
.Convert<Gray, byte>()
.PyrDown()
.PyrUp()
.Not()
.InRange(new Gray(190), new Gray(255))
.Canny(new Gray(190), new Gray(255))
.FindContours(CHAIN_APPROX_METHOD.CV_CHAIN_APPROX_SIMPLE,
RETR_TYPE.CV_RETR_TREE);
var output = new Image<Gray, byte>(orig.Size);
for (; contours != null; contours = contours.HNext)
{
var poly = contours.ApproxPoly(contours.Perimeter*0.05,
contours.Storage);
output.Draw(poly, new Gray(255), 1);
}
output.Save(outFile);
This is the result:
Approach 2
Convert the image to grayscale
Remove noise and invert
Remove anything that is not really bright
Get the canny and then lines
Code for this:
var orig = new Image<Bgr, byte>(inFile);
var linesegs = orig
.Convert<Gray, byte>()
.PyrDown()
.PyrUp()
.Not()
.InRange(new Gray(190), new Gray(255))
.Canny(new Gray(190), new Gray(255))
.HoughLinesBinary(
1,
Math.PI/45.0,
20,
30,
10
)[0];
var output = new Image<Gray, byte>(orig.Size);
foreach (var l in linesegs)
{
output.Draw(l, new Gray(255), 1);
}
output.Save(outFile);
This is the result:
Notes
I have tried adjusting all the parameters on those two approaches and adding smoothing but I can never get the simple edges that I need because, I suppose, the darker region is not a solid colour.
I have also tried dilating and eroding but the parameters I have to put in for those are so high to get a single colour that I end up including some of the grey stuff on the right and lose accuracy.

Yes, it's possible, and here is how you could do it:
Change the contrast of the image to make the lighter part disappear:
Then, convert it to HSV to perform a threshold operation on the Saturation channel:
And execute erode & dilate operations to get rid of the noises:
At this point you'll have the result you were looking for. For testing purposes, at the end I execute the bounding box technique to show how to detect the beggining and the end of the area of interest:
I didn't have the time to tweak the parameters and make a perfect detection, but I'm sure you can figure it out. This answer provides a roadmap for achieving that!
This is the C++ code I came up with, I trust you are capable of converting it to C#:
#include <cv.h>
#include <highgui.h>
int main(int argc, char* argv[])
{
cv::Mat image = cv::imread(argv[1]);
cv::Mat new_image = cv::Mat::zeros(image.size(), image.type());
/* Change contrast: new_image(i,j) = alpha*image(i,j) + beta */
double alpha = 1.8; // [1.0-3.0]
int beta = 100; // [0-100]
for (int y = 0; y < image.rows; y++)
{
for (int x = 0; x < image.cols; x++)
{
for (int c = 0; c < 3; c++)
{
new_image.at<cv::Vec3b>(y,x)[c] =
cv::saturate_cast<uchar>(alpha * (image.at<cv::Vec3b>(y,x)[c]) + beta);
}
}
}
cv::imshow("contrast", new_image);
/* Convert RGB Mat into HSV color space */
cv::Mat hsv;
cv::cvtColor(new_image, hsv, CV_BGR2HSV);
std::vector<cv::Mat> v;
cv::split(hsv,v);
// Perform threshold on the S channel of hSv
int thres = 15;
cv::threshold(v[1], v[1], thres, 255, cv::THRESH_BINARY_INV);
cv::imshow("saturation", v[1]);
/* Erode & Dilate */
int erosion_size = 6;
cv::Mat element = cv::getStructuringElement(cv::MORPH_CROSS,
cv::Size(2 * erosion_size + 1, 2 * erosion_size + 1),
cv::Point(erosion_size, erosion_size) );
cv::erode(v[1], v[1], element);
cv::dilate(v[1], v[1], element);
cv::imshow("binary", v[1]);
/* Bounding box */
// Invert colors
cv::bitwise_not(v[1], v[1]);
// Store the set of points in the image before assembling the bounding box
std::vector<cv::Point> points;
cv::Mat_<uchar>::iterator it = v[1].begin<uchar>();
cv::Mat_<uchar>::iterator end = v[1].end<uchar>();
for (; it != end; ++it)
{
if (*it) points.push_back(it.pos());
}
// Compute minimal bounding box
cv::RotatedRect box = cv::minAreaRect(cv::Mat(points));
// Draw bounding box in the original image (debug purposes)
cv::Point2f vertices[4];
box.points(vertices);
for (int i = 0; i < 4; ++i)
{
cv::line(image, vertices[i], vertices[(i + 1) % 4], cv::Scalar(0, 255, 0), 2, CV_AA);
}
cv::imshow("box", image);
cvWaitKey(0);
return 0;
}

Image doesn't update when written to.. weird goings on

I have a Kinect WPF Application that takes images from the Kinect, does some feature detection using EmguCV (A C# opencv wrapper) and displays the output on the using a WPF image.
I have had this working before, but the application now refuses to update the screen image when the imagesource is written to, but I have not changed the way it works.
the Image(called video) is written to as such:
video.Source = bitmapsource;
in the colorframeready event handler.
This works fine until I introduce some opencv code before the imagesource is written to. It does not matter what source is used, so I don't think it is a conflict there. I have narrowed down the offending EmguCV code to this line:
RecentKeyPoints = surfCPU.DetectKeyPointsRaw(ImageRecent, null);
which jumps straight into the opencv code. It is worth noting that:
ImageRecent has completely different origins to the bitmapsource updating the screen.
Reading video.Source returns the bitmapsource, so it seems to be writing correctly, just not updating the screen.
Let me know if you want any more information...
void nui_ColorFrameReady(object sender, ColorImageFrameReadyEventArgs e)
{
// Checks for a recent Depth Image
if (!TrackingReady) return;
// Stores image
using (ColorImageFrame colorImageFrame = e.OpenColorImageFrame())
{
if (colorImageFrame != null)
{
if (FeatureTracker.ColourImageRecent == null)
//allocate the first time
FeatureTracker.ColourImageRecent = new byte[colorImageFrame.PixelDataLength];
colorImageFrame.CopyPixelDataTo(FeatureTracker.ColourImageRecent);
}
else return;
}
FeatureTracker.FeatureDetect(nui);
//video.Source = FeatureTracker.ColourImageRecent.ToBitmapSource();
video.Source = ((Bitmap)Bitmap.FromFile("test1.png")).ToBitmapSource();
TrackingReady = false;
}
public Bitmap FeatureDetect(KinectSensor nui)
{
byte[] ColourClone = new byte[ColourImageRecent.Length];
Array.Copy(ColourImageRecent, ColourClone, ColourImageRecent.Length);
Bitmap test = (Bitmap)Bitmap.FromFile("test1.png");
test.RotateFlip(RotateFlipType.RotateNoneFlipY);
Image<Gray, Byte> ImageRecent = new Image<Gray, byte>(test);
SURFDetector surfCPU = new SURFDetector(2000, false);
VectorOfKeyPoint RecentKeyPoints;
Matrix<int> indices;
Matrix<float> dist;
Matrix<byte> mask;
bool MatchFailed = false;
// extract SURF features from the object image
RecentKeyPoints = surfCPU.DetectKeyPointsRaw(ImageRecent, null);
//Matrix<float> RecentDescriptors = surfCPU.ComputeDescriptorsRaw(ImageRecent, null, RecentKeyPoints);
//MKeyPoint[] RecentPoints = RecentKeyPoints.ToArray();
// don't feature detect on first attempt, just store image details for next attempt
#region
/*
if (KeyPointsOld == null)
{
KeyPointsOld = RecentKeyPoints;
PointsOld = RecentPoints;
DescriptorsOld = RecentDescriptors;
return ImageRecent.ToBitmap();
}
*/
#endregion
// Attempt to match points to their nearest neighbour
#region
/*
BruteForceMatcher SURFmatcher = new BruteForceMatcher(BruteForceMatcher.DistanceType.L2F32);
SURFmatcher.Add(RecentDescriptors);
int k = 5;
indices = new Matrix<int>(DescriptorsOld.Rows, k);
dist = new Matrix<float>(DescriptorsOld.Rows, k);
*/
// Match features, provide the top k matches
//SURFmatcher.KnnMatch(DescriptorsOld, indices, dist, k, null);
// Create mask and set to allow all features
//mask = new Matrix<byte>(dist.Rows, 1);
//mask.SetValue(255);
#endregion
//Features2DTracker.VoteForUniqueness(dist, 0.8, mask);
// Check number of good maches and for error and end matching if true
#region
//int nonZeroCount = CvInvoke.cvCountNonZero(mask);
//if (nonZeroCount < 5) MatchFailed = true;
/*
try
{
nonZeroCount = Features2DTracker.VoteForSizeAndOrientation(RecentKeyPoints, KeyPointsOld, indices, mask, 1.5, 20);
}
catch (SystemException)
{
MatchFailed = true;
}
if (nonZeroCount < 5) MatchFailed = true;
if (MatchFailed)
{
return ImageRecent.ToBitmap();
}
*/
#endregion
//DepthMapColourCoordsRecent = CreateDepthMap(nui, DepthImageRecent);
//PointDist[] FeatureDistances = DistanceToFeature(indices, mask, RecentPoints);
//Image<Rgb,Byte> rgbimage = ImageRecent.Convert<Rgb, Byte>();
//rgbimage = DrawPoints(FeatureDistances, rgbimage);
// Store recent image data for next feature detect.
//KeyPointsOld = RecentKeyPoints;
//PointsOld = RecentPoints;
//DescriptorsOld = RecentDescriptors;
//CreateDepthMap(nui, iva);
//rgbimage = CreateDepthImage(DepthMapColourCoordsRecent, rgbimage);
// Convert image back to a bitmap
count++;
//Bitmap bitmap3 = rgbimage.ToBitmap();
//bitmapstore = bitmap3;
//bitmap3.Save("test" + count.ToString() + ".png");
return null;
}

This is a little late, but I had a similar problem and thought I'd share my solution.
In my case I was processing the depth stream. The default resolution was 640x480, and Emgu just wasn't able to process the image fast enough to keep up with the frameready handler. As soon as I reduced the depth stream resolution to 320x240 the problem went away.
I also went a bit further and moved my image processing to a different thread which sped it up even more (do a search for ComponentDispatcher.ThreadIdle). I'm still not able to do 640x480 at a reasonable frame rate, but at least the image renders so I can see what's going on.