Pixel is a 2D or 3D? please answer me. thanks

Pixel is a 2D or 3D? please answer me. thanks

Answer 1

[Image Analyst]

A grayscale, indexed, or binary image is 2D. A true color, RGB image is 3D. Just look at how many values you get if you specify two values (row and column) as the "independent variables".

[Muhammad Sohail]

Thanks Image Analyst

[Muhammad Sohail]

As i am asking about individual Pixel is that answer also applied to the pixel too? i-e pixel in gray scale is 2D and in True RGB its 3D?

[Stephen23]

This answer is misleading, because a pixel is not the same as the data that represents it.

A pixel is, by definition, a picture element: "is a physical point in a raster image".

Given that there are many possible ways of encoding one pixel, there could be many possible dimensions of the data used to encode one pixel. The data of one pixel from an indexed image has a different number of dimensions to an RGB or RGBA or some other more complicated format, yet they could all represent the same pixel when it is displayed as an image.

A voxel is truly 3D, which implies that a pixel exists only in a 2D plane (and not the number of dimensions of the data used to encode it).

[Image Analyst]

Well, yeah, there's a little more involved for a truly complete answer. Kind of like how people call something created by surf() a 3D plot, or call an x-y plot a 2D plot - there's more to it. I was thinking about the type of variable that would be used to hold the image.

[Muhammad Sohail]

Thanks stephen, but still i dont have clear answer of that whether a pixel can have any dimension or not? if it has any dimension what will be the context for that? and what dimension in that context it have.. :(

[Stephen23]

@Muhammad Sohail: Don't confuse a pixel (a point on an image) with the data that is used to define that pixel.

I think you are actually asking about how many dimensions the pixel data has, in which case the dimensions of that data depends on how the image is encoded (it might be 2D, 3D, etc, as Image Analyst explained). The only way to know is to read about that image format, and check that particular image. Here is a good introduction to common ways of encoding/storing image data:

http://www.mathworks.com/help/matlab/creating_plots/image-types.html

You can see on that page that depending on how the image is encoded the data that encodes a pixel may be 2D or 3D (or other dimensions), but by definition they all encode a single pixel (point of that image). The indexed image uses a 2D matrix, while the RGB image uses a 3D matrix, yet the pixels of the image are exactly the same in the example images.

A pixel is strictly defined as a point/element of a 2D image. An element of a 3D image (such as used in VR and games) is called a voxel, not a pixel.

Answer 2

[Mike Garrity]

I think that the best answer for what a pixel "is" is Alvy Ray Smith's A Pixel Is Not A Little Square memo.

[Image Analyst]

I don't buy everything Alvy says there. A CCD "well" is a square (for the vast majority of sensors, Fuji being an exception), which means that when mapped back onto the scene, the signal is the mean signal from a square back in the original scene. True, that square from the scene may not be a square upon display, like on an old trinitron monitor or a dithered printout, but mathematically the signal still represents light coming from a square patch back in the original scene that reflected light into the camera.

[Mike Garrity]

Yes but ... In most cases a CCD well doesn't end up being a pixel because demosaicing is performed to combine values from multiple wells behind different color filters to create the final pixel.

http://www.dpreview.com/glossary/camera-system/color-filter-array

There's usually an antialiasing filter in there too. These have the effect of smearing some of the light from the edges of the well into the neighboring well to prevent artifacts from features which are smaller than the width of the well.

And then on display, antialiasing schemes such as ClearType take advantage of the internal structure of the "pixel" to increase the apparent resolution.

There are cases where you can think of a pixel as being a little square, but there are also a lot of cases where that mental model is going to get you into trouble. I think that Alvy was correct to tell us not to get too comfortable with that mental model.

[Image Analyst]

True, the birefringent anti-aliasing filter over the sensor does give a duplicate image that makes the square a bit sheared. I thought demosaicing was just done on the "missing" pixels, and there of course it's just an interpolation, or estimate, of what the value would be there in the missing spot. So, yes, 2/3 of the pixels will not be exactly a square with really straight, sharp, crisp edges. But on the pixels that are actually there, don't they just deliver those unaltered (not demosaiced and affected by neighboring pixels)?

And of course cameras might add their own secret sauce to the final image they deliver such as dead pixel removal, various color enhancements (some of which you can specify/control and some that you have no control over), and perhaps edge enhancements or other special effects.

Answer 3

[Thorsten]

A pixel is a "picture element", or the basic element of a digital image. The most common image formats are binary images, where the pixel can be either 0 or 1, gray scale images where the pixel can take values 0, 1, 2,..., 255 (8bit) or RGB color images, where the pixel is defined by three 8-bit values for Red, Green, and Blue, respectively.

[Muhammad Sohail]

Thanks Thorsten it really help me, as am new to this field so different questions raising in mind... :)