For designers working in graphic design companies using Adobe Photoshop CS4, it is important to understand the various resolution-attributes like dimensions, bit depths, and color models. These are discussed below:
The dimension of an image should be set with a proper pixel value. Although, Photoshop has a maximum pixel limit of 300000 X 300000, graphics designers usually prefer to keep the dimension to a lower resolution (for example, 400X400 pixels). Similarly, one can work with bit depths up to 32 floating-point bits per channel in HDR (High Dynamic Range) mode in graphic web design ; although, most image editing is done at 8 bits per channel (24-bit RGB and 32-bit CMYK).
The different resolution types are:
Sampling Resolution – For scanned images, sampling resolution gives a measurement of precision of the scanning process which is measured in sample per inch (spi).
Document Resolution – The resolution value saved in the file when an image is imported from any software, is known as document resolution. The image quality changes only when the image is resampled.
Effective resolution – The effective resolution is the resolution of the image after it is scaled (by compressing or stretching the pixels) to its final print size.
Device resolution – It is the resolution of the output device. In graphic & web design, where the output device is not specific, it is essential to keep the resolution to a common and acceptable figure.
To give details of the resolution of an image, following parameters are important:
Pixel Dimensions Physical Dimensions and Resolution File Size Megapixels Single-Side Dimension. ‘Res’ or the number of pixels per millimeter
A bigger resolution does not necessarily mean a better resolution. A higher means that it would take longer to get saved, printed or opened. At the same time, a lower resolution may result in a pixelated image with a degraded quality. So, it is essential for a web graphic designer to keep the right resolution balance depending on the purpose of the final image.
I am the webmaster at www.synapse-consultants.com ? a custom web development company offering numerous services, such as content management, offshore software development, online marketing, search engine optimization, search marketing, and website maintenance services.

I've seen multidimensional scaling used for handwriting recognition, but the similarity measure there comes from eigenanalysis of a pixelated image of the actual text. In principal I suppose the same thing could be done with digitized books. I'm not sure how the data from these type of captchas could inform that process.
I'm pretty unfamiliar with neural nets, do they work similarly to that?