Distance of objects from webcam

> Hi, i wish to find the distance from the webcam to different objects in
> an image. can this be done in some way?
> Joseph
>
One way is to use markers that your system would recognize or findalready marked objects (objects with visible distinctive and easy tofind features). You need to calibrate your camera first so as toincorporate all the distortions of the lens in your distancecalculation. Then you use FindExtrinsicCameraParams2 to find thelocation and orientation of your target. The target must consist ofat least 4 distinctive points of known relative location. This wouldbe the input of the function (the relative location of the 4 pointsand their location in the image). Not to mention that it also needsintrinsic_matrix, distortion_coeffs but these you will obtain duringcalibration.I have been using this method with a simple target of four squaresand it works nicely.
Have a look here: http://www.youtube.com/watch?v=4OK8ROW77pc

Message from OpenCV

Locate License Plate

[OpenCV] ANPR/Car number-plate recognition

You can use bottomhat transform to get rid of trouble in case of the same colorof car and plate. In addition, you can use radon transform to correct skewing. Itried so many methods to divide characters from the plate candidate region.First I extract the exact plate rectangle from the plate candidate region. ThenI apply global threshold and labeling to get characters as objects. I elect thenoncharacter object by some rules (area, width/length etc.). This method hassome trouble while the exact plate region is somehow connected to the otherregions. Does anyone have suggestion?

Alper

License Plate Localization

License Plate Localization

You should look gabor filter. You have to find best filter parametersfor detection plate character, and than you can optain all platesafter filtering. Than you can use standart morp. operations on theimage.

https://commerce.metapress.com/content/0f1a8wk9mcru0djd/resource-secured/?target=fulltext.pdf&sid=sswqda45miyw0i55iwfxto45&sh=www.springerlink.com

License Plate Localization

[OpenCV] License Plate Localization

After detecting edges, you should apply morphological close operation. However,you should select kernel size considering the plate size. After that, you shouldapply connected component algorithm and filter very big and very small sizeobjects. Hence, the remaining objects are plate candidates which you can detectcharacters for each candidate.

Alper

Merging 2 Image

Hi,
a simple solution is:
You could initialize a new picture with the required dimensions, andcopy the data line for line, one after one from both pictures into thenew one.

Examplecode:
//Let's assume img1 and img2 are your 1024x768 grayscale-images of
// type IplImage

IplImage *combined = new IplImage;
CvSize size;
size.width=2048;
size.height=768;
combined = cvCreateImage(size, IPL_DEPTH_8U, 1);
for (int i = 0; i < (size.height-1); i++) { memcpy(&img3->imageData[i*img3->widthStep],&img1->imageData[i*img1->widthStep], img1->widthStep);
memcpy(&img3->imageData[i*img3->widthStep+img1->widthStep],&img2->imageData[i*img2->widthStep], img2->widthStep);
}