What is... a scanner?
The interface TWAIN
Although it is a question of a topic that compete definitely to the software (the programs), it does not come badly to mention briefly of what this element consists. It is a question of a norm that was defined so that any scanner could be used by any program of a standardized form and enclosed with the same interface for the acquisition of the image.
Although a few years ago there still existed a relatively high number of devices that were using other proper methods, nowadays it is possible to be said that all the normal escáners use this protocol, with what the manufacturers owe only to worry about providing the controler appropriate TWAIN, generally in versions for Windows 9x, NT and sometimes 3.x. Unfortunately, only the relatively expensive mark escáners bring controlers for other operating systems as OS/2 or Linux, and even sometimes not even for Windows 3.x or NT; the good news is that the TWAIN standardization does that sometimes we could use the controler of another scanner of similar characteristics, although obviously it is not a desirable method...
Making separate the bookstores DLL and other technical topics, the part that the user sees of the standard TWAIN is the interface of acquisition of images. It is a question of a program in which of a visual form we can control all the parameters of the scanning (resolution, number of colors, sheen...), in addition to being able to define the size of the area that we want to process.
If the loyalty of the color is an importance factor, one of the parameters that probably we will have to modify in this one interface is the gamma control, to fit the scale of colors that receives the scanner with the one that presents our monitor or prints the printer. I recommend to him to look in the papers of the scanner for more information.
It is a question of one of the most common applications of the escáners. OCR is the initials of Optical Character Recognition, optical recognition of characters, or with a simpler description: how to do to teach to read to the computer.
If we think a little about the process of scanning that we have described previously, we will realize that, on having scanned a text, letters, words and phrases are not scanned, but simply the points that form them, a species of photo of the text. Obviously, this can be useful to file texts, but it would be desirable that we could take all these so interesting but so heavy references and incorporate them into our text processor not like an image, but like text editable.
What we would wish finally would be that the computer could read like us. Very well, since that does the OCR: it is a program that reads these digital images and looks for sets of points that are alike letters, to characters. Depending on the complexity of the above mentioned program it will understand more or less types, going so far as in some cases to interpret the manual writing, support the original format (columns, photos between the text...) or to apply grammatical rules to increase the accuracy of the process of recognition.
So that the program could realize these tasks with a certain reliability, without confusing "t" with "1", for example, the image that we provide to him it must fulfill a few certain characteristics. Fundamentally it must have a big resolution, approximately 300 ppp for texts with clear types or 600 ppp if it is a question of small or original types of few quality like newspapers. On contrary, we can save in the aspect of the color: almost always there will be enough white person and black (1 bit of color), or at most a scale of gray 256 (8 bits). For this motive some roller escáners (much adapted for this type of tasks) lack support for color.
The necessary team
We cannot finish this explanation on the escáners without giving a few notions about the essential hardware them to handle (and the advisable one, that as it is always superior to this one... and more expensive).
At present, thanks to the file formats with compression the storage of the images is completely solved, especially in this epoch of hard disks of 4 GB and more; the unsolvable problem is the memory necessary RAM to present the images on the screen and to treat them or to print them.
Since we have said in several occasions, it is not strange that an image occupies in memory 25 MB or any more; therefore, at the moment when we overcome the memory physically installed in the computer (nowadays approximately 64 MB, although it even does well small were 16 MB or even less) the computer will make use of the virtual memory, which is not but memory imitated thanks to the hard disk. The problem is that this "false memory" is many thousands of slower than times the RAM, what can eternize the process, in addition to putting at the edge of his resistance the very unstable Windows 95 (and his older brother the quite unstable one 98).
For all this, to work with a scanner it turns out to be essential to have at least 16 MB of RAM, being absolutely necessary to come up to 32 MB if we are usually going to be employed with originals at color in formats that overcome them 10x15 cm. And if our target happens for scanning images A4 or bigger than high resolutions (600x600 ppp or more) and then to treat them in the computer (for example for desktop publishing, digital printing works or to spend to electronic format planes of architecture or engineering), the absolute minimum are 64 MB.
Otherwise, the computer does not need a few high services; 486 can be enough with a microprocessor, although bearing in mind that the digital images treatment is a process that makes use of enough any potency increase in this sense. Be reasonable, if a scanner for 12.000 pts is bought he will not need a Pentium III, but if he buys it to himself of 50.000 pts do not tune in to a computer that costs less than the scanner...