I am not a software engineer. Excuse me if you find the question awkward.
I'd like to have an image format which is not supposed to be memory efficient but easy to manipulate in plain C. To be more specific, I desire to store every pixel in an array of the form:
pixel[row#][column#][Color]
where the indexes row# and column# (255 at max) are coordinates, and the index Color (2 at max) contains the RGB values of the pixel specified by the position ( i.e. pixel[255][255][1] is used to check or manipulate the Green amount inside the pixel on the bottom right corner ).
I aim to use this form in robotic applications to be able to find the coordinates of the first red/blue/green pixel easily by scanning the image starting from the top left corner using "nested for loops" (yes, not a creative solution). Here, you might say, if there is a white area on the image, the code will return with wrong coordinates. I am aware of this fact but the images will not have a complex pattern, and (if necessary) i can store the irrelevant colors as if they were black. I do not care about the brightness, gamma, alpha whatever, too.
So is it possible to write a C (or C++ if mandatory) code to take snapshots from the webcam say at every 0.5 seconds and convert the raw image from the webcam to the form specified above? If you say C can not reach to the camera directly, is it possible to write a code which calls for a software that can reach the camera, take a snapshot and then store the raw data in a file? If yes, how can i read this raw data file using C codes to be able to at least try a conversion? I am using Windows Vista on my laptop.
Sorry for keeping the question long, but I don't want to leave any points unclear.
Yes, such a file format would be possible. Only sanity prevents it from being implemented/used.
Some formats in fairly wide use would be almost as simple to scan in the way you're considering though. A 32-bit BMP, for one example, has a small header on the file to tell a few things like the size of the picture (x,y pixel dimensions) followed by the raw pixel values, so it's basically just ColorColorColor... for the number of pixels in the image.
Code to do the scanning you're talking about with a 32-bit BMP will be pretty trivial -- the code to open the file, allocate space for a buffer to read data into, etc., could easily be longer than the scanning code itself.
Adopting a 'standard' image format also means you have tools to generate test data and independently view your results
The easiest to code in pure C (even if it's not very efficent) is portable pixmap (ppm).
It's just a plain text file
P3 <cr> # P3 means an ascii color file R,G,B
640 480 <cr> # 640 pixels wide, 480 rows deep
255 <cr> # maximum value is 255
# then just a row of RGB values separated by space with a CR
255 0 0 0 255 0 0 0 255 #....... 640 triplets <cr>
255 255 0 255 255 255 0 0 0 # ....... next row etc
There is also a more efficnet binary version where the data is pure RGB bytes so is very easy to read into a C array in a single operation
I am not a software engineer. Excuse me if you find the question awkward.
I'd like to have an image format which is not supposed to be memory efficient but easy to manipulate in plain C. To be more specific, I desire to store every pixel in an array of the form:
pixel[row#][column#][Color]
where the indexes row# and column# (255 at max) are coordinates, and the index Color (2 at max) contains the RGB values of the pixel specified by the position ( i.e. pixel[255][255][1] is used to check or manipulate the Green amount inside the pixel on the bottom right corner ).
I aim to use this form in robotic applications to be able to find the coordinates of the first red/blue/green pixel easily by scanning the image starting from the top left corner using "nested for loops" (yes, not a creative solution). Here, you might say, if there is a white area on the image, the code will return with wrong coordinates. I am aware of this fact but the images will not have a complex pattern, and (if necessary) i can store the irrelevant colors as if they were black. I do not care about the brightness, gamma, alpha whatever, too.
So is it possible to write a C (or C++ if mandatory) code to take snapshots from the webcam say at every 0.5 seconds and convert the raw image from the webcam to the form specified above? If you say C can not reach to the camera directly, is it possible to write a code which calls for a software that can reach the camera, take a snapshot and then store the raw data in a file? If yes, how can i read this raw data file using C codes to be able to at least try a conversion? I am using Windows Vista on my laptop.
Sorry for keeping the question long, but I don't want to leave any points unclear.
Yes, such a file format would be possible. Only sanity prevents it from being implemented/used.
Some formats in fairly wide use would be almost as simple to scan in the way you're considering though. A 32-bit BMP, for one example, has a small header on the file to tell a few things like the size of the picture (x,y pixel dimensions) followed by the raw pixel values, so it's basically just ColorColorColor... for the number of pixels in the image.
Code to do the scanning you're talking about with a 32-bit BMP will be pretty trivial -- the code to open the file, allocate space for a buffer to read data into, etc., could easily be longer than the scanning code itself.
Adopting a 'standard' image format also means you have tools to generate test data and independently view your results
The easiest to code in pure C (even if it's not very efficent) is portable pixmap (ppm).
It's just a plain text file
P3 <cr> # P3 means an ascii color file R,G,B
640 480 <cr> # 640 pixels wide, 480 rows deep
255 <cr> # maximum value is 255
# then just a row of RGB values separated by space with a CR
255 0 0 0 255 0 0 0 255 #....... 640 triplets <cr>
255 255 0 255 255 255 0 0 0 # ....... next row etc
There is also a more efficnet binary version where the data is pure RGB bytes so is very easy to read into a C array in a single operation
0 commentaires:
Enregistrer un commentaire