I'm studying SuperResolution techniques, and I read often "subpixel"/"subpixel registration"/"subpixel accuracy". I don't find what exactly it means; Wikipedia says that is one of the three RGB channel in the pixel, but I think it isn't the right meaning in this context. Anyone can help explain what it is it or give me any tutorial?
$\begingroup$
$\endgroup$
1
-
$\begingroup$ I'm lost that if we have two pixels v1 and v2 as mentioned as top,and their positions are recpectively p1(10,3) and p2(33,7). So how to calculate the p(avg). How to apply the formula p(avg)=(p1v1+p2v2)/(v1+v2); $\endgroup$– EL HAMDAOUI AtmanJan 10, 2017 at 11:20
Add a comment
|
3 Answers
$\begingroup$
$\endgroup$
Images are technically limited to the original pixel resolution. Objects in an image are spatially quantized at this resolution. However, an edge for instance may be localized "between" native pixels. Traditional tools like basic gradients often stick to that initial resolution. However, using additional information (multiple data, motion vectors, prior), one can infer a more accurate object location with a sub-pixel resolution, i.e. with an accuracy finer than that of the pixel width, at a fraction of the pixel size, with non-integer coordinates.
This can also apply to track/detect objects with a size smaller than that of a pixel. See for instance: What is a subpixel?
Using clever subpixel shifts, one can for instance reveal asymmetry not visible in the original image, as shown below:
Old and recent references:
- Image Sequence Enhancement Using Sub-pixel Displacements, 1988
- Multi-frame image super-resolution reconstruction using sparse co-occurrence prior and sub-pixel registration, 2013
Note: subpixels is also used in subpixel rendering on displays, and refers to colored pixel components (red, green, blue, plus yellow, white or cyan) that are displayed, or viewed, like a single pixel.
answered Sep 5, 2016 at 12:47
$\begingroup$
$\endgroup$
2
Sub-pixel means that instead of getting a location in the image in terms of $(x,y)$ coordinates at the pixel level, which means integer values for $x$ and $y$, the location is calculated to possibly give fractional pixel locations.
For example, instead of getting $(10,12)$ as a location, the algorithm may give $(9.934, 12.1234)$.
Suppose you have two pixels with values $v_1$ and $v_2$ and their pixel positions are $p_1$ and $p_2$ (which are integers). The average pixel position calculation could be: $$ p_{\tt avg} = \frac{p_1 v_1 + p_2 v_2}{v_1 + v_2} $$ which will, in general, not give an integer value for $p_{\tt avg}$.
-
1
-
$\begingroup$
$\endgroup$
Sub-pixel is a virtual pixel defined between two actual pixels in an image capturing sensor. to enhance the resolution or image quality, sub-pixel calculation is very useful. https://baike.baidu.com/item/%E4%BA%9A%E5%83%8F%E7%B4%A0/7904986?fr=aladdin
