SucesiÃ³n apostÃ³lica

One indicates under the term of digital image any image (Dessin, icon, Photographie…) acquired , created , treated or stored in binary form (continuation of 0 and 1):

Acquired by devices like the scanners, the cameras or numerical video cameras, the charts of video acquisition (which directly digitize a source like the Télévision).
Created directly by computer programs, via the mouse, the graphics tablets or by the Modeling 3D (what one calls by abuse language the “synthesized images”).
Treated thanks to computer tools. It is easy to modify it in the face, color, to add or remove elements, to apply varied filters, etc
Stockée to a data-processing support (Disquette, Hard drive, CD-ROM,…)

Types of images

One distinguishes two types from images to the composition and the different behavior: matric images and the vectorial images.

Matric images (or images bitmap)

It is made up as its name indicates it of a matrix (table) of points to several dimensions, each dimension representing a space dimension (height, width, depth), temporal (duration) or other (for example, a level of resolution).

Images 2D

In the case of the images with two dimensions (more running), the points are called Pixel S. From a mathematical point of view, one considers the image as a function of

\ mathbb R \ times \ mathbb R

\ mathbb R

where the verse of entry is regarded as a space position, the singleton of exit like a coding.

This type of image adapts well to posting on data-processing screen (he also directed pixel); it on the other hand is adapted little for the impression, because the resolution of the data-processing screens, generally from 72 to 96 PPP (“points per inch”, in English dowries per inch or dpi ) is quite lower than that reached by the printers, at least 600 PPP today. The printed image, if it does not have an high-resolution, will be thus more or less fuzzy or will let appear visible square pixels.

Images 2D + T (video), images 3D, images multi-resolution

When an image has a temporal component, one speaks about Animation.
Danslecasde the images with three dimensions the points are called Voxel S . They represent a Volume.

These cases are a generalization of the case 2D, additional dimension representing time respectively, a space dimension or a scale of resolution.

From a mathematical point of view, it is about a function of $\ mathbb R \ times \ mathbb R \ times \ mathbb R$ in $\ mathbb R$

Stereoscopic images

It is about a particular case in which one works by couples of images, the latter being able to be of any of the preceding types.

There exists a great number of kinds of images stereoscopic, and a even greater number of means to observe them in relief, but the coding recommended by the international organizations of Stéréoscopie is indicated like " jps" , i.e. a format jpg in which the two sights left and right-hand side are juxtaposed in the same file, generally 2048x768, each of the two sights being registered in a rectangle 1024x768 and, if his width ratio/height is not 4/3, each sight is supplemented in this rectangle by two symmetrical black bands, either in top and bottom, or on the left and on the right.

Vectorial images

The principle is to represent the data of the image by geometrical formulas which could be described from a mathematical point of view. That means that instead of memorizing a mosaic of elementary points, one stores the succession of operations leading to the layout. For example, a drawing can be memorized by the computer like “ a straight line plotted between the points (x1, y1) and (x2, y2) ”, then “ a traced circle of center (x3, y3) and of ray 30 of red color ”.

The advantage of this type of image is the possibility indefinitely of increasing it without losing initial quality, as well as a compactness. The use of predilection of this type of images relates to the diagrams which it is possible to generate with certain software of CAD (Drawing Computer-assisted) like AutoCAD or CATIA. This type of images is also used for animations Flash, used on Internet for the creation of advertizing banners, the introduction of Web sites, even of the complete Web sites.

Since the current means of visualization of images as the monitors of computer rest primarily on matric images, the vectorial descriptions (Files) must beforehand be converted into matric descriptions before being posted like images.

Definition and resolution

The matric images are also defined by their definition and them resolution.

The definition of an image is defined by the number of points composing it. In digital image, that corresponds to the number of pixels which composes the image in height (vertical axis) and width (horizontal axis): 200 pixels by 450 pixels for example, shortened in “200×450”.

The resolution of an image is defined by a number of pixels per unit of length of the structure to digitize (classically in PPP). This parameter is defined at the time of the Numérisation (passage of the image in binary form ), and depends mainly on the characteristics of the material used at the time of digitalization. The higher the number of pixels per unit of length of the structure to be digitized is, the more the quantity of information which describes this structure is important and the higher the resolution is. The resolution of an digital image defines the degree of detail of the image. Thus, more the resolution is high, better is the restitution.

However, for the same dimension of image, plus the resolution is high, plus the number of pixels composing the image is large. The number of pixels is proportional to the square of the resolution, being given the two-dimensional character of the image: if the resolution is multiplied by two, the number of pixels is multiplied by four. To increase the resolution can involve times of longer visualization and impression, and lead to a too important size of the file containing the image and to occupied excessive place in memory.

Representation of the colors

There exist several modes of data-processing Codage of the colors, more used for the handling of the images is the colorimetric Espace Rouge, Vert, Bleu (RVB or RGB - Red green Blue). This space is based on a additive Synthèse colors, i.e. the mixture of the three components R, V, and B with their maximum value gives white, following the example Lumière. The mixture of these three colors to various proportions makes it possible to reproduce with the screen a big part of the Visible spectrum, without having to specify a multitude of luminous frequencies.

There exist other modes of representation of the colors:

Cyan, Magenta, Yellow, Black (CMJN or CMYK) used mainly for the impression, and based on a subtractive Synthesis of the colors;
Dyed, Saturation, Brightness (TSL or HSL), where the color is coded according to the circle of the colors;
bases optimal color YUV, Y representing brightness, U and V two orthogonal Chrominance S.

The images bitmap colors can of which to be represented either by an image in which the value of the pixel is a linear combination of the values of the three components colors, or by three images representing each one a component color. In the first case, according to the number of bits (elementary unit of information which can take two distinct values) allocated for the storage of a color of pixel, one generally distinguishes the various following types of images:

Images 24 bits (or “true colors”)

It is about a misleading name because the numerical world (finished, limited) cannot give an account of reality completely (infinite). The coding of the color is carried out on three Octet S, each byte representing the value of a component color by an entirety from 0 to 255. These three values generally code the color in space RVB. The number of different colors which can be thus represented is of 256 X 256 X 256 possibilities, that is to say nearly 16 million colors. As the difference in nuance between two colors very close but different in this mode from representation is almost unperceivable for the human eye, one considers conveniently that this system allows an exact restitution of the colors, this is why one speaks about “true colors”.

The images bitmap based on this representation can quickly occupy a considerable storage space, each pixel requiring three bytes to code its color.

Images with pallets, images 256 colors (8 bits)

To reduce the place occupied by the information of color, one uses a palette “attached” to the image. One speaks then about indexed colors: the value associated with a pixel does not convey any more the effective color of the pixel, but returns at the entry corresponding to this value in a table (or pallets) colors called lookup counts or LUTE in English, in whom one has the complete representation of the color considered.

According to the number of colors present in the image, one can thus gain a considerable place: it is considered in practice that 256 colors among the 16 million colors 24 bits are sufficient. To code them, there will be thus a pallet occupying 24 bits X 256 entries, that is to say 3 X 256 bytes, and the pixels of the image will be associated with indices coded on a byte. The occupation of such an image is thus of 1 byte by pixel plus LUTE, which represents a little more of the third of the place occupied by an image colors 24 bits (the more the image contains pixels, the more the space saver is important, the limit being one the third of the place occupied by the image true colors).

Another existing method consists in doing without pallet, and to directly code the three colors by using a byte: each component color is coded on two bits, the bit remaining can be used either to manage more colors on one of the components, or to manage the transparency of the pixel. With this method, one obtains images bitmap with a coding color actually limited to 8 bits, although the beach of the possible colors is very reduced compared to that which the method offers using a pallet.

In the case of the images indexed colors, it is possible to specify that the pixels using one of the colors of the pallet are not posted during the reading of the data of the image. This property of transparency is very much used (and useful) for the images of the Web pages, so that the prime coat of the image does not prevent the visualization of the background of the page.

Images in colors (or levels) of gray

One codes here nothing any more but the level of the luminous intensity, generally on a byte (256 values). By convention, zero value represents the black (null luminous intensity) and value 255 the white (maximum luminous intensity):

This process is frequently used to reproduce photographs in black and white or text under certain conditions (with use of a filter to soften contours in order to obtain smoother characters).

This coding of the simple luminous intensity is also used for the coding of images colors: the image is represented by three images of light intensity, each one being in a component distinct from colorimetric space (for example, intensity of red, green and blue).

Images with management of translucidity

One can allot to an image an additional channel, called channel alpha , which defines the degree of transparency of the image. It is about a channel similar to the traditional channels defining the components of color, coded on a fixed number of bits per pixel (in general 8 or 16). One thus linearly spreads out the translucidity of a pixel, complete opacity with the transparency.

Other formats

Certain original formats were used:

the Mode HAM on Amiga is a format famous for its rather special coding of the colors. Indeed, the color of a pixel was described using that of the pixel immediately on its left of which one of the components of color was modified or drawn from a pallet of 16 colors.
the Mode Halfbrite always on Amiga which made it possible to circumvent a limitation of the Amiga pallets which were limited to 32 colors. 32 other colors were deduced from the first by decreasing their intensity of half.

Formats of images

Definition

A format of image is a representation Informatique of the image, associated with Information S on the way in which the image is coded and possibly providing indications on the manner of decoding it and of handling it.

Structuring, use of Metadatum S

Precautions to be taken

Some precautions to be taken concerning the formats of images:

the formats known as “owners”, can differ according to the software which handles them. Moreover, their perenniality is not guaranteed: to realize new programs to read them can to prove difficult (especially if their specifications were not made public), that can even if to prove illegal if the algorithms used are protected by Brevet S.
It is necessary to pay attention to the various versions which can cover a particular format. In particular for the format tiff which varies according to the versions; some of them are not recognized by certain software.

Comparative table

Formats owners

Format tiff is regarded as a format owner, the patent being controlled by the firm Aldus.

In the past, format GIF was subjected to the Unisys patent controlled by the company CompuServe, it was thus a format owner. But it should be noted that the patents of Unisys expired on June 20th, 2003 at the United States, on June 18th, 2004 in the majority of the countries of Europe, on June 20th, 2004 in Japan and on July 7th, 2004 in Canada. It thus became since a free format of rights.

Digital image and royalty

To try to make respect the Royalty (in France) and the Copyright (in almost all the other countries), there exist techniques of numerical marking of an image. These techniques, which one names impressed, are used more and more. The print is supposed to preserve a proof of the origin of the image, in the shape of a visible or invisible signature, which must resist the treatments likely to be applied to the image. This “tattooing” can be done according to two methods, generally indicated by the same term of filigree :

Protection by visible signature

This technique consists in integrating an indication on the image, for example the organization or the author at which belongs the image, in order to dissuade the pirates to make use of it. The disadvantage of this method is that it is very easy to eliminate this type of tattooing with a tool for image processing, since tattooing is visible.

Protection by encrypted signature

This technique consists in hiding tattooing in the data of the image. This approach with the advantage of not obstructing the reading of the image by the simple spectator while allowing an easy identification. The author draws a complementary advantage from it: the possible inattentive pirate will not be tried to withdraw or modify the signature; the more voluntary pirate will see his illegal activity made a little more difficult or easily provable (by the only presence of tattooing).

See also: Stéganographie

Random links: