Opencv 2.4.0: Load image through unsigned char array - arrays

I think I've thoroughly searched the forums, unless I left out certain keywords in my search string, so forgive me if I've missed a post. I am currently using OpenCV 2.4.0 and I have what I think is just a simple problem:
I am trying to take in an unsigned character array (8 bit, 3 channel) that I get from another API and put that into an OpenCV matrix to then view it. However, all that displays is an image of the correct size but a completely uniform gray. This is the same color you see when you specify the incorrect Mat name to be displayed.
Have consulted:
Convert a string of bytes to cv::mat (uses a string inside of array) and
opencv create mat from camera data (what I thought was a BINGO!, but can't seem to get to display the image properly).
I took a step back and just tried making a sample array (to eliminate the other part that supplies this array):
int main() {
bool isCamera = true;
unsigned char image_data[] = {255,0,0,255,0,0,255,0,0,255,0,0,255,0,0,255,0,0,0,255,0,0,255,0,0,255,0,0,255,0,0,255,0,0,255,0,0,0,255,0,0,255,0,0,255,0,0,255,0,0,255,0,0,255};
cv::Mat image_as_mat(Size(6,3),CV_8UC3,image_data);
namedWindow("DisplayVector2",CV_WINDOW_AUTOSIZE);
imshow("DisplayVector2",image_as_mat);
cout << image_as_mat << endl;
getchar();
}
So I am just creating a 6x3 matrix, with the first row being red pixels, the second row being green pixels, and third row being blue. However this still results in the same blank gray image but of correct size.
The output of the matrix is (note the semicolons i.e. it formatted it correctly):
[255, 0, 0, 255, 0, 0, 255, 0, 0, 255, 0, 0, 255, 0, 0, 255, 0, 0; 0, 255, 0, 0, 255, 0, 0, 255, 0, 0, 255, 0, 0, 255, 0, 0, 255, 0; 0, 0, 255, 0, 0, 255, 0, 0, 255, 0, 0, 255, 0, 0, 255, 0, 0, 255]
I might be crazy or missing something obvious here. Do I need to initialize something in the Mat to allow it to display properly? Much appreciated as always for all your help everyone!

all the voodoo here boils down to calling getchar() instead of (the required) waitKey()
let me explain, waitKey might be a misnomer here, but you actually need it, as the code inside wraps the window's messageloop, which triggers the actual blitting (besides waiting for keypresses).
if you don't call it, your window will never get updated, and will just stay grey ( that's what you observe here )
indeed, you should have trusted the result from cout , your Mat got properly constructed, it just did not show up in the namedWindow
(btw, getchar() waits for a keypress from the console window, not your img-window)
hope it helps, happy hacking further on ;)

Related

glTexImage2D, in random and occasional ways, expends 800 times more in executing

I'm using glsl 2.0 for some GPGPU purposes (I know, not the best for GPGPU).
I have a reduction phase for matrix multiplication in which I have to constantly reduce the texture size (I'm using glTexImage2D). The pseudocode is something like this:
// Start reduction
for (int i = 1; i <= it; i++)
{
glViewport(0, 0, x, y);
glDrawArrays(GL_TRIANGLES, 0, 6);
x = resize(it);
if (i % 2 != 0)
{
glUniform1i(tex2_multiply_initialstep, 4);
glBindFramebuffer(GL_FRAMEBUFFER, framebuffer3);
// Resize output texture
glActiveTexture(GL_TEXTURE5);
glTexImage2D(GL_TEXTURE_2D, 0, GL_RGBA32F, x, y, 0, GL_RGBA, GL_FLOAT, NULL);
}
else
{
glUniform1i(tex2_multiply_initialstep, 5);
glBindFramebuffer(GL_FRAMEBUFFER, framebuffer2);
// Resize output texture
glActiveTexture(GL_TEXTURE4);
// A LOT OF TIME!!!!!!!
glTexImage2D(GL_TEXTURE_2D, 0, GL_RGBA32F, x, y, 0, GL_RGBA, GL_FLOAT, NULL);
// A LOT OF TIME!!!!!!!
}
}
In some iterations the glTexImage2D of the else branch takes 800 times more time that in other ones. I make a test hardcoding x and y but surprisingly takes similar high times in the same iterations, so have nothing to do with x value.
What's wrong here? Alternatives to resizing without glTexImage2D?
Thanks.
EDIT:
I know that glsl 2.0 is a bad choice for GPGPU but its mandatory for my project. So that I'm not able to use functions like glTexStorage2D because they are not included in 2.0 subset.
I'm not sure if I understand exactly what you're trying to achieve, but glTexImage2D is reallocating memory each time you call it. You may want to call glTexStorage2D, and then call glTexSubImage2D.
You can check Khronos's Common Mistakes page about that. Relevant part is :
Better code would be to use texture storage functions (if you have
OpenGL 4.2 or ARB_texture_storage) to allocate the texture's storage,
then upload with glTexSubImage2D:
glGenTextures(1, &textureID);
glBindTexture(GL_TEXTURE_2D, textureID);
glTexStorage2D(GL_TEXTURE_2D, 1, GL_RGBA8, width, height);
glTexSubImage2D(GL_TEXTURE_2D, 0​, 0, 0, width​, height​, GL_BGRA, GL_UNSIGNED_BYTE, pixels);
This creates a texture with a single mipmap level, and sets all of the parameters appropriately. If you wanted to have multiple mipmaps, then you should change the 1 to the number of mipmaps you want. You will also need separate glTexSubImage2D calls to upload each mipmap.
If that is unavailable, you can get a similar effect from this code:
glGenTextures(1, &textureID);
glBindTexture(GL_TEXTURE_2D, textureID);
glTexParameteri(GL_TEXTURE_2D, GL_TEXTURE_BASE_LEVEL, 0);
glTexParameteri(GL_TEXTURE_2D, GL_TEXTURE_MAX_LEVEL, 0);
glTexImage2D(GL_TEXTURE_2D, 0, GL_RGBA8, width, height, 0, GL_BGRA, GL_UNSIGNED_BYTE, pixels);
Again, if you use more than one mipmaps, you should change the GL_TEXTURE_MAX_LEVEL to state how many you will use (minus 1. The base/max level is a closed range), then perform a glTexImage2D (note the lack of "Sub") for each mipmap.

Arduino Binary Array is too Large

I have a three-dimensional array of binary numbers, which I use as a dictionary and pass through an LED array. The dictionary covers 27 letters, and each letter covers 30x30 pixels (where each pixel is a 0 or a 1).
I was using the Intel Edison - and the code worked well - but I ditched the Edison after having trouble connecting it to my PC (despite replacing it once). I switched to the Arduino Uno, but am now receiving an error that the array is too large.
Right now I have the array set as boolean. Is there anyway to reduce the memory demands of the array by setting it instead as bits? The array consists of just zeros and ones.
Here's a snip of the code:
boolean PHDict[27][30][30] = {
/* A */ {{ 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 1, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0}, /* this is one column of thirty, that show "A" as a letter */
You could write it as
#include <stdint.h>
//...
uint32_t PHdict[27][30] = {
{ 0x00004000, ... },
....
};
.. Where each entry contains 30 bits packed into a 32-bit number.
The size is under 4k bytes.
You would need a bit of code to unpack the bits when reading the array, and a way to generate the packed values (I.e a program which runs on your "host" computer, and generates the initialized array for the source code)
For the AVR processor, there's also a way to tell the compiler you want the array stored in PM (Flash memory) instead of DM - I think if you have it in DM, the compiler will need to put the initialization data in PM anyway, and copy it over before the program starts, so it's a good idea to explicitly store it in PM. See https://gcc.gnu.org/onlinedocs/gcc/AVR-Variable-Attributes.html#AVR-Variable-Attributes
In fact, depending on the amount of flash memory in the processor, changing it to PM may be sufficient to solve the problem, without needing to pack the bits.

Smooth Operation In Image Processing

I would like to optimize my code,
My code contains following steps
//img size 1024 X 720
cvSmooth(img, inputImg, CV_GAUSSIAN, 5, 0, 0);
FilterOperations::sobelSchaar(inputImg, sobImg);
cvResize(sobImg, sob2Img);
//sob2Img size 512 X 360
cvSmooth(sob2Img, sob2Img, CV_GAUSSIAN, 3, 3);
imgProceObj.normalise(sob2Img);
cvAdaptiveThreshold(sob2Img, sob2Img, 255,
CV_ADAPTIVE_THRESH_GAUSSIAN_C, CV_THRESH_BINARY, 31, 50);
How can I optimize above code, without changing final result ?
pyrDown() might be doing both required steps in one go.

Color Depth PIXELFORMATDESCRIPTOR

I'm wondering what values to change in a PIXELFORMATDESCRIPTOR object to change the color depth.
According to the OpenGL wiki, this is how you'd create a PIXELFORMATDESCRIPTOR object for an OpenGL context:
PIXELFORMATDESCRIPTOR pfd =
{
sizeof(PIXELFORMATDESCRIPTOR),
1,
PFD_DRAW_TO_WINDOW | PFD_SUPPORT_OPENGL | PFD_DOUBLEBUFFER, //Flags
PFD_TYPE_RGBA, //The kind of framebuffer. RGBA or palette.
32, //Colordepth of the framebuffer.
0, 0, 0, 0, 0, 0,
0,
0,
0,
0, 0, 0, 0,
24, //Number of bits for the depthbuffer
8, //Number of bits for the stencilbuffer
0, //Number of Aux buffers in the framebuffer.
PFD_MAIN_PLANE,
0,
0, 0, 0
};
But it has different variables effecting the color depth.
Which ones do I need to change to adjust the color depth accordingly?
The first number, 32 in your particular example specifies the amount of color bitplanes available to the framebuffer. The other numbers define the numbers of bitplanes to use for each component. It's perfectly possible to fit a 5-6-5 pixelformat into a 32 bitplanes framebuffer, which is a valid choice.
When you pass a PIXELFORMATDESCRIPTOR to ChoosePixelFormat the values are takes as minimum values. However the algorithm used by ChoosePixelFormat may not deliver an optimal result for your desired application. It can then be better to enumerate all available pixelformats and choose from them using a custom set of rules.

gdk_pixbuf_composite usage

I have two png images First one with Width1 2247 Height1 190 and second one with Width2 155 Height2 36. I wan't the second image(src) to be placed in the center of first image(dest). I created pixel buf of both and used gdk_pixbuf_composite as follows.
gdk_pixbuf_composite( srcpixbuf, dstpixbuf, 1000, 100, width2, height2, 0, 0, 1, 1, GDK_INTERP_BILINEAR, 255);
I get a hazy window of width2 and height2 on the first image.
If I replace width2 and height2 with 1.0 then I don't get the srcimage on the dstimage. Where am I going wrong?
gdk_pixbuf_composite( srcpixbuf, dstpixbuf, 1000, 100, width2, height2, 1000, 100, 1, 1, GDK_INTERP_BILINEAR, 255);
This solved. Wrongly understood the offset parameter. Basically an intermediate scaled image is created and only the part represented by the dest wid, height is composited. So in my case we need to move the entire unscaled image to the destination offset which is done by the offset parameter.

Resources