What is... a scanner?
Abiding by the criteria of the Real Academy of the Language, famous by the brilliant introduction of the term cederrón to name to the CD - ROM, probably not at all; for the rest of common mortal, let's say that it is the word that is used in computer science to designate to a device image digitizer.
For digitizing the operation is understood of transforming something analogical (something physical, real, of infinite precision) into something digital (a finite set and of certain precision of logical units named bits). Anyway, that leaving of so much syntactic formalism to us, in the case that occupies us it is a question of taking an image (photo, drawing or text) and of converting it to a format that we could store and of modifying with the computer. Really a scanner is not not any more nor less than the eyes of the computer.
How does it work
The process of reception of an image turns out to be almost identical for any scanner: the image is illuminated with a light focus, one leads by means of mirrors the light reflected towards a device named CCD that transforms the light into electrical signs, it transforms the above mentioned electrical signs to digital format into a DAC analogical - digital (converter) and the resultant bits wealth is transmitted to the computer.
The CCD (Charge Coupled Device, device connected by load - electricity company-) is the fundamental element of any scanner, independently of his form, size or mechanics. It consists of an electronic element that reacts to the light, transmitting more or less electricity as it is the intensity and the color of the light that it receives; it is an authentic electronic eye. Nowadays it is quite common, it is possible that you possess one without knowing it: in his video camera, in his fax, in his digital camera of photos...
The final quality of the scanning will depend fundamentally on the quality of the CCD; other elements will be able to do a better or worse work, but if the image is not received by loyalty any later operation will not be able to arrange the problem. Bearing in mind the previous thing, also we must bear in mind the quality of the DAC, since of nothing it serves to receive the light with enormous precision if we lose a lot of of this information on having transformed the electrical wealth to bits.
For this motive it usually say that there are preferable the escáners of marks of prestige as Nikon or Kodak to others with a major theoretical resolution, but with CCDs that they do not receive with loyalty the colors or DACs who do not make use well of the electrical sign, giving poorer, flatter results.
We cannot continue the explanation without defining this term, one of the parameters most used (sometimes even too much) at the time of determining the quality of a scanner. The resolution (measurement in ppp, points by inch) can be defined as the number of individual points of an image that is capable of receiving a scanner... although in fact it is not anything so simple.
The this way definite resolution would be the optical or real resolution of the scanner. This way, when we speak about a scanner with resolution of "300x600 ppp" we are referring that in every horizontal line one inch long (2,54 cm) can receive 300 individual points, while in vertical it comes up to 600 points; as in this case, generally the horizontal resolution and the vertical one do not coincide, being major (typically the double) the vertical one.
This optical resolution is given by the CCD and is the most important, since it implies the physical limits of quality that we can obtain with the scanner. For it, it is a very typical commercial method to comment only on the biggest of two values, describing like "a scanner of 600 ppp" to a device of 300x600 ppp or "a scanner of 1.200 ppp" to a device of 600x1.200 ppp; bear it in mind, the difference is to obtain or not the points fourfold one.
We have also the interpolated resolution; it consists of overcoming the limits that there imposes the optical resolution (300x600 ppp, for example) what might be the values of the points that we add for software to the image by means of the mathematical estimation. For example, if the scanner receives physically two contiguous points, the white one and another black, he will suppose that of c have received an extra point between both it would be of some gray tone. From this form we can come to absurdly high resolutions, of even 9.600x9.600 ppp, although in fact we do not obtain any more real information than that it provides the maximum optical resolution of the device. Obviously this value is the one that more the escáners advertisers like...
Finally the proper scanning resolution is, that one that we select to receive a concrete image. His value will go from a certain minimum (typically approximately 75 ppp) up to the maximum of the interpolated resolution. In this case the value is always identical for the horizontal resolution and the vertical one, since if not the image would have the deformed dimensions.
The colors and the bits
On having spoken about images, digitalises or not, there escapes from nobody the importance that has the color. A color photograph turns out to be much more agreeable of seeing that other one in gray tones; a graph correctly colored turns out to be much more interesting than black and white other; even a text in which the epigraphs or the conclusions have an out-standing color turns out to be less monotonous and invites to his reading.
Nevertheless, to digitize the infinite tones that can exist in a photo any is not a simple process. Till recently, the escáners were receiving the images only black and white or, as much, with a much limited number of tones of gray, between 16 and 256. Later it appeared escáners that could receive color, although the process needed three spent ones over the image, one for every primary color (red, blue and green). Nowadays the practical totality of the escáners they receive up to 16,7 million different colors in the past only one, and even some they come up to 68.719 million colors.
To understand how it goes over to these resounding numbers we must tell how the colors assign the computers to the images. In all the computers there uses what it is named binary system, which is a mathematical system in which the top unit is not 10 as in the decimal system to which we are accustomed, but 2. An any bit can take therefore 2 values, which can represent colors (white person and black, for example); if instead of a bit we have 8, the possible values are 2 risen up to 8 = 256 colors; if they are 16 bits, 2 risen up to 16 = 65.536 colors; if they are 24 bits, 2 risen up to 24 = 16.777216 colors; etc, etc.
Therefore, "an image to 24 color bits" is an image in which every point can have up to 16,7 million different colors; this colors quantity is considered to be sufficient for almost all the normal uses of an image, for what he usually name a real color. Almost totality of the current escáners they capture the images with 24 bits, but the current tendency consists of scanning even with more bits, 30 ó even 36, in such a way that there is received a bogey of colors absolutely faithful to the real one; nevertheless, almost always this color depth comes down later to 24 bits to support a size of reasonable memory, but the final quality keeps on being very high since only the most redundant color information is eliminated.
How much occupies an image?
It depends on the image (brilliant answer: truth?). To know exactly which is going to be the size of an image, we will have to use the following formula:
Size image (KB) = L x A x RH x RV x bits / 8.192
Where L and A are the dimensions of the image in inches (one inch = 2,54 cm) and RH and RV the horizontal and vertical resolutions respectively. Let's do a rapid example: an image DIN-A4 (approximately 11,7x8,3 inches) scanned ppp (300x300) with 24 bits of color (real color) occupies 300: 25.490 KB!! (approximately 25 MB, 25 gentle ones!!). The number proves impactante, but do not worry; many methods exist to reduce the size of the images, so much at the time of handling them in memory as to that of storing them on the hard disk.
The first method consists of scanning to minor resolution; the quality is minor, but the size of the resultant file also. If the image is going to take the screen of a computer as a destination, 75 ppp will be almost always sufficient, what would reduce the size of the image previous to scarcely 1.593 KB, little more than 1,5 MB.
As the second method we have to reduce the color depth. If the previous image is a drawing to Chinese ink, in spite of scanning to 1 (black and white) bit it is possible that we have sufficiently. This would limit the size to only 1.062 KB, almost exactly 1 MB.
Finally we can file the image in compressed format. In this case the size of the image in memory remains invariable (25 MB), but the size on disc can stay in less of fifth one part without quality loss, or even less if the compression is realized eliminating redundant information. As example of formats of file of image with compression we have the JPEG (or JPG), GIF or TIFF, opposite to the outstanding figure BMP who lacks some compression.
The most important thing is that we can combine the previous factors to manage turned out really optimized; this way, scanning the image of the example to 75 ppp, with 1 color bit and keeping it in format GIF, the result can occupy only 66 KB in memory and less than 15 KB on disc.
To end with this topic we are going to put a table they sum up in that there learns the quantity of memory RAM that some typical examples of original occupy to different resolutions and colors:
It is necessary to emphasize that in many cases there are used scales of gray 256 (8 bits) to represent more faithfully black and white originals with very definite rims or small sizes of letter. On what is it the OCR we will treat at the end of the article, although if he is in the hurry to know it it pulsates here.