WO2007063195A1

WO2007063195A1 - Method of coding in graphic form

Info

Publication number: WO2007063195A1
Application number: PCT/FR2005/003013
Authority: WO
Inventors: Oguz Berke Durak
Original assignee: Universite Paris 7 Denis Diderot
Priority date: 2005-12-02
Filing date: 2005-12-02
Publication date: 2007-06-07

Abstract

An item of information is coded in the form of a graphic mark, with several connected components. For coding, the information item to be coded is transformed into a tree. Thereafter, the tree is transformed into a graphic mark; a node of the tree is coded by a connected component of the graphic mark with as many interior connected components as the node has children. For decoding, the connected components of the graphic mark are considered and the graphic mark is transformed into a tree; a node of the tree corresponds to a connected component of the graphic mark and the node of the tree has as many children as the corresponding connected component has interior connected components. The tree is thereafter transformed into information. The graphic mark obtained in accordance with the coding is decodable, even after deformations: recognition of connected components is resistant to the deformations of the graphic mark.

Description

GRAPHIC UNDERFORMING CODING METHOD

The present invention relates to a method of encoding information in graphical form and a code obtainable by such a method. It also relates to a method of decoding a code in graphical form.

Information coding methods in graphical form have been proposed, in which information is coded in graphic form. In the remainder of the description, the following terms are used:

- the word "information" represents the information to be coded; it is for example a number, base 10 or binary representation;

the words "coding in graphic form" or "coding" represent the method making it possible to transform the information into a representation in graphic form, adapted if necessary to be read by a machine;

the word "graphic code" or "graphic mark" represents the result of the coding in graphic form, that is to say a set of marks readable by a reading device;

the word "decoding" represents the inverse method of the coding, making it possible to transform a graphic code into the corresponding information.

It is understood that the simple writing of a number in the form of numbers or even of letters represents, in the strict sense, a coding in graphical form. This coding is adapted to a human reading; it is not or poorly suited to reading by a machine, since the recognition of numbers and letters requires relatively complex software tools. The present invention applies to encodings adapted to a reading by a machine. Among the coding methods in graphical form, a commonly used method is bar code coding. The graphic code called "bar code" or "bar code" is composed of a plurality of parallel bars. The respective widths of the bars and / or their relative positions are representative of the information. The bar code is adapted to be read by a scanning device using a laser or an image sensor. Barcodes are commonly used for logistical purposes or for the identification of products in the trade. This method is for example described in the EAN product identification standards, available in France from Gencod EAN France. An example of a bar code of the state of the art is shown in FIG. 1; under the bars of the code is reproduced, in the usual way, the information. This coding method has the disadvantage that the code is sensitive to a deformation of the support on which it is applied. Thus, an extension of the support in a direction perpendicular to the direction of the bars can change the value of the code or make the code unreadable. It is also necessary to have a good reading resolution in the direction perpendicular to the bars of the code.

More elaborate versions of such codes are described in US-A-5,504,322 or US-A-6,047,892.

Coding methods providing two-dimensional graphic codes have been proposed; these codes are based on the definition of cells in the graphic code; these cells take values representative of the symbols to be coded. For example, EP-A-I 443 452 proposes a two-dimensional coding using three colors, with reference areas. US-A-2003/0121978 proposes a positioning of the cells in the code which renders the reading more robust. US-A-2005/0061892 contains a description of different symbols usable for two-dimensional codes. US-A-2005/0127191 provides encoding using image processing algorithms for decoding two-dimensional graphic codes. US-A-2001/0007116 proposes a distribution of the symbols in the graphic code as well as a redundancy allowing a better resistance to the transmission errors, in particular by fax. US-A-2004/0031852 proposes to have in the code a synchronization pattern. US-A-6,267,296 provides a two-dimensional code with symbols of different sizes; redundancy allows a correct reading under different conditions.

US-A-6,360,948 mentions the problem of the distortion caused by the reading of an image on a non-plane support; he suggests to have in the image alignment symbols to take account of this deformation; the proposed system aims to compensate for the form of the medium on which the code is applied.

Other solutions of the state of the art are described in documents FR-A-2 809 210 (coding of the 256 possible values of one byte by different symbols), FR-A-2 735 891 (reading of the pixels of an image), FR-A-2 660 773 (redundant coding), FR-A-2 561 012 (method and device for pattern recognition), EP-AI 287 483, WO-A-2004 055735, WO-A A-01 86582 (coding using separate areas of reference, data, control and parity), or WO-A-03 012727 (coding using symbols consisting of angular orientations of the modulation patterns of an image) WO-A-02 080 645 (recognition of marks). The solutions described in these documents consist in choosing symbols that are supposed to be more resistant to errors or in arranging in the graphic code elements facilitating the recognition of specific areas of the code. These solutions are, in each case, adapted to a given type of error. The graphic codes of these documents are hardly or hardly readable in case of deformation of the medium on which the graphic code is applied - curling of the support, inhomogeneous stretching, folding; they are as little or hardly readable in the presence of deformations due to the reading devices; for example, deformations due to camera lenses, perspective effects or curl surfaces make it difficult to read these codes. There is therefore a need for a graphic code, which supports the deformation, whether it is the deformation of the medium on which the code is applied or the deformation caused by the reading of the graphic code. Such a graphic code is all the more useful that it is possible to code and decode it using simple means.

Accordingly, the invention provides, in one embodiment, a method of encoding information in the form of a graphical mark having a plurality of related components, comprising:

a step of transforming the information to be coded into a tree;

a step of transforming the tree into a graphic mark, in which a node of the tree is encoded by a connected component of the graphic mark comprising as many internal connected components as the present node of son. The method may have one or more of the following additional features:

the tree is a commutative or non-commutative tree;

- the tree is a binary tree; the connected components have at least two colors, and a node of the tree is encoded by a connected component of the graphic mark having a first color and comprising as many internal connected components of a second color as the present node of wires.

In another embodiment, the invention provides a method of decoding a graphical mark having a plurality of related components into an information; the method comprises:

a step of transforming the graphic mark into a tree, in which a node of the tree corresponds to a connected component of the mark graph the node with as many wires as the corresponding connected component has inner connected components;

a step of transforming the tree into information.

As in the coding method, it can be provided that the tree is a commutative, non-commutative tree or a binary tree.

If the related components have at least two colors, a node of the tree corresponding to a connected component of the graphical mark having a first color has as many threads as the related component comprises inner connected components of a second color. In another embodiment, the invention proposes a support having a graphic mark capable of being decoded according to this method; yet another embodiment relates to a support with a graphic mark obtained by the method described above.

An indication of the value of the information decoded or coded according to said method, for example a directly human-readable indication, may be provided on the medium. A related component of the graphic mark may have a rectangular shape.

Advantageously, the support has at least one of the following characteristics:

the tree corresponding to the mark in the coding or the decoding has a height greater than or equal to 5;

the tree corresponding to the mark in the coding or the decoding has a number of nodes greater than 10

- the tree corresponding to the mark in the encoding or decoding satisfies

max _\ 2, p) 'with N the number of nodes, p the maximum arity of the nodes of the tree and h the height of the tree;

- the diameter of the largest uniform color disc in the mark is less than or equal to four times the line thickness used to write the mark

- density, expressed as the ratio between

the quantity of coded information, measured in bits and

the ratio of the area of the graphic mark and the square of the line thickness is greater than 0.5%. In yet another embodiment, the invention provides a decoder apparatus, comprising:

an acquisition part adapted to provide an image of a graphic code; - A processing part that can be performed by software or cable and adapted to receive the image provided by the acquisition part and to decode the received image to provide the value of the information coded in the graphic code. The processing portion may also be adapted to provide an indication of the position of a graphic code in the received image.

Finally, the invention provides a computer program on a computer readable medium and containing code adapted to execute on a computer all the steps of the encoding or decoding method mentioned above.

Other characteristics and advantages of the invention will appear on reading the following description of embodiments, given by way of example only and with reference to the appended drawings, in which:

- Figure 1 shows a bar code of the state of the art;

FIG. 2 is a flow chart of an embodiment of the method of the invention; FIG. 3 shows a tree representative of the information "12";

FIG. 4 shows a graphic code encoding the information "12";

FIG. 5 shows another graphic code encoding information "12";

FIG. 6 is a flow chart of a decoding algorithm of an embodiment of the method of the invention; FIG. 7 shows a graphic code encoding the same information as the bar code of FIG. 1;

FIGS. 8 to 12 show different examples of graphic codes encoding the "777" information;

FIG. 13 shows a non-commutative tree representative of the information "12", in a second embodiment of the invention;

FIG. 14 shows a graphic code coding information "12";

FIG. 15 shows another graphic code coding information "12";

- Figure 16 shows a decoding apparatus according to the invention.

The invention proposes, in one embodiment, to use, in a graphic code, the connected components to graphically represent the information to be coded. Specifically, the information is recursively encoded by the number of related components contained in a given connected component.

In this context, we call "connected component" a set satisfying the following property: for two elements a, b of the set, there exists a continuous path allowing to go from the element a to the element b passing only by elements of the set. This definition of connectivity corresponds to the topology definition of arc connectivity.

In a digitized graphic mark, formed of a plurality of pixels having for example two values (black or white), a component (black for example) is connected if any pixel of the component can be connected to another pixel of the component by a path formed of adjacent pixels of the component. It is understood that while this definition may depend on the resolution of the digitization, it is independent of the reason for the digitization and the definition of the adjacent one; thus, a conventional digitization in the form of a two-dimensional matrix of pixels leads to the same result, with regard to the connectivity, as a digitization with a matrix of hexagonal elements.

The use of the related components, instead of symbols of a predetermined shape or position, ensures that the code is readable despite deformations.

The code is particularly resistant to inhomogeneous extensions - stretching in one or more directions -, curl or deformation of the support. The code still resists deformations that can be made by a reading device.

FIGS. 3 to 5 give a first example of an embodiment of the invention. In this first embodiment, a representation of the information in the form of a commutative tree is used. This solution avoids having to have an orientation of the graphic code.

In addition, a graphic code is used in which the connected components are alternately black and white; this choice ensures a better compactness of the graphic code; this choice is independent of the commutative or non-chosen representation for the tree.

In this first example, the information to be coded is the number twelve ("12"). This number is written in binary 1100. The first step 12 of the flow chart of FIG. 2 consists in choosing the information to be coded.

For coding, the information is transformed into a tree, which is represented in the second step 14 of the flow chart of FIG. 2. Any algorithm known per se can be used for this purpose. In the example, we considered the following algorithm. Let w be a binary word of length m representing the information to be coded. Its bits are numbered from 1 to m, the z ^' -th bit being denoted w (i). The noncommutative binary trees are represented by a set of terms t satisfying the equation t = A + B (t) + C (t, t). These terms are constructed as follows:

1. The tree reduced to a single node is represented by ^. 2. The term B (t) represents the tree whose root has only one son, this son being the root of the tree represented by t.

3. The term C (t, t) represents a tree whose root has two sons, the first being the root of the tree represented by t and the second that of the tree represented by f. Here is a recursive procedure T transforming every word w into a term T (w). When m =

0 we put T (W) = A. Otherwise let w (l) be the first bit of w and w '= w (2) ... w (m) the word consisting of the last m - 1 bits of w. Let x (respectively y) be the word obtained by deleting the digits in the even (odd) positions of w '. If w (l) = 0 we put T (w) = B (T (w ')), otherwise we put T (w) = C (T (x), T (y)). The Injective Test transformation and the inverse transform are easily deduced from this definition. For example, the word w = 1100 is transformed according to this method into

T (w) = C (C (B (A) J) β (A)))

Here are two recursive procedures U and V transforming terms into terms. The procedure U makes it possible to transform a term t into a term U (t) so that for any term t 'obtained by rearrangement of the order of the threads of the nodes of U (t) we have V (f) = t .

In other words, U makes it possible to transform an noncommutative binary tree into a commutative binary tree in an injective manner, and V makes it possible to retrieve the non-commutative tree from its transformed transform. This is important because the information on the order of the threads is lost when the read symbol undergoes deformations, rotations or reflections, and allows a greater freedom in the procedure of graphic formatting. We put U (A) = A, V (A) = A and for all terms t and t ':

U (B (ή) = B (U (ή)

V (B (t) = B (V (t))

U (C (t, t ')) = C (B (U (t), C (Aβ (U (f))))) V (C (B (t), C (Aβ (t)))) ≈ V (C (B (t), C (B (t) A)))

V (C (A, B (t) \ B (t))) ≈ V (C (B (t), C (B (t) A))) V (C (C (B (f) A ) β (i))) = C (V (t), V (t '))

Note that the function V is partial, the cases not covered by the definition correspond to trees that were not obtained by the procedure U. The term t = T (OOIl) of the preceding example is transformed into

U (C (C (B (A)) β (A)))) ≈ C (B (C (B (B (A) ₅ C (A ₅ B (A)))), C (A ₅ B (B (A))))) The results obtained by the application of this algorithm to the coding of the number 12 are shown in FIG. 3. For clarity in the presentation, the nodes of the tree of FIG. 3 are numbered from 40 to 54. The node 40 is the root node. The nodes are associated with references A, B and C, as indicated above. The tree is commutative, in that a change in the order of the threads is not important for the transformation of information into a tree and for the inverse transformation of the tree into information. Thus, exchanging the relative position of the nodes 41 and 42 (and their respective progenies) does not change the coding.

In addition, the tree in Figure 3 shows that the nodes are represented in black or white, the color of a node depending on its depth in the tree - the nodes having an even depth are black while the nodes with an odd depth are white. It is only a convention of representation to facilitate the understanding of the presentation, as shown by the comparison of Figures 4 and 5 below.

In the next step of the flowchart, referenced 16, the root node 40 is considered as the current node.

In the next step 18, a black connected component is drawn for the current node, having as many internal connected components as the current node having son nodes. At the first iteration, the current node is the root node 40, with two child nodes. We trace a connected component, which contains two related components. These are respectively in Figures 4 and 5 of the outermost connected components, referenced 40a and 40b.

In the next step 20, it is sought if there is another node of the same depth in the tree. If this is the case, change the next node to another node of the same depth

(step 22) and go back to step 18; if there is no other node of the same depth, we go to step 24. In the example, at the first iteration, there is no node of the same depth as the root node and we pass directly at step 24.

In step 24, it is checked whether there are nodes of lower level; if this is the case, go to step 26, and if not go to step 28. In the example, there are nodes of a depth greater than the root node and so we go to step 26 In step 26, change the current node to a lower level node and loop to step 18. In the example, and go from the root node to node 41 or node 42, for example to the node. 41. Then, loop to step 18.

In step 28, the drawing of the graphic code is completed. In the example, beginning with the root node, it is thus necessary to draw successively the following connected components 40a to 54a in FIG. 4 corresponding to the nodes 40 to 54 of FIG. 3. The connected components 53a and 54a are not strictly speaking "drawn", since they are implicitly defined when drawing the connected components 50a and 52a.

Of course, the flow chart of FIG. 2 is only an example of implementation of a coding according to the first embodiment. In this example, the nodes of the tree of FIG. 3 are scanned in the ascending order of the references of the figure. It is obvious that the same graphic code is obtained by scanning the nodes of the tree in another order. In the same way, the algorithm proposed above for the transformation of information into a tree is only an example; it has the advantage of simplicity and efficiency, but one could also consider another algorithm.

In the example of Figure 4, we considered connected components having a shape of rectangle or square. This solution has the advantage of simplicity. More precisely, nested rectangles with parallel edges are used.

Figure 5 shows another example of graphic code, also corresponding to the number "12". This example shows that the invention can provide graphical marks easily readable by a machine, but can nevertheless be made in a form that can be appreciated by the human being independently of the coded information. In the example of FIG. 5, the different graphic components are referenced 40b to 54b and correspond respectively to the nodes 40 to 54 of the tree of FIG. 3. Compared with FIG. 4, the black and white nodes have been inverted. the outer frame of FIG. 5 does not constitute a useful connected component in the coding, but simply the border of the graphic code. The connected component corresponding to the root node 40 is the white connected component 40b surrounding the bicycle; it contains two connected components 41b and 42b, corresponding to the nodes 41 and 42. It can be seen that from a topological point of view, FIG. 5 is identical to FIG. 4 - only changing the shape of the various connected components forming the object, as well as the alternation white-black, as indicated above.

In the example of Figure 4 as in that of Figure 5, the size of the graphic code can be determined in advance by counting the depth of the tree. This makes it possible to determine the size of the graphic code.

A decoding algorithm corresponding to the coding algorithm of FIG. 3 is described with reference to FIG. 6. The first step of the algorithm consists in creating a tree from the received graphic code; iteratively, each connected component of the code graph corresponds to a node of the tree whose number of wires is a function of the number of related components contained in this component. The second step of the algorithm is to transform the tree into information.

The example in Figure 6 gives more details on the first step; in step 112, the graphics code is received; this reception of the graphic code is carried out by image acquisition, in a manner known per se. A charge-coupled camera, a scanning device (scanner) or any other solution for acquiring an image in the form of pixels can be used. As noted above, the resolution may vary, since the resolution is sufficient to preserve the connectivity properties in the image. In practice, it is sufficient that the picture element is smaller (in its largest dimension) than half the minimum distance between two connected components; this criterion ensures that the acquisition of the image of the graphic code preserves the connectivity of the different components of the code. As indicated above, the grid used may be square, hexagonal, or irregular; an irregular grid is particularly suitable if the graphic code is deformed and this deformation is predetermined or measurable on the code - for example, for a code with components formed of lines of constant thickness, the variations in thickness of the lines may be used as a measure of deformation.

Conventional image processing operations can be applied to the image of the graphic code, such as linear or morphological filter denoising, followed by thresholding. In this way we obtain an image having a finite number of colors (two in the examples) and having homogeneous areas.

Next, in step 114, the first connected component - the outermost connected component - is considered and defined as the current connected component. The current node of the tree is defined as the root node. Next, step 116 is performed. At step 116, the number of connected components in the current connected component is determined and one or more wires are constructed at the current node based on the number of related components contained within. of the current connected component. Each of the son nodes thus constructed is associated with a connected component. Next, step 118 is performed. In step 118, it is determined whether there are corresponding connected components in the tree at nodes of the same depth as the current node. If this is the case, proceed to step 120, in which the current connected component is changed to the identified connected component and the current node to the corresponding node, then to step 116. Conversely, if it is determined in step 118 that there are no related components corresponding in the tree to nodes of the same depth as the current node, go to step 122.

In step 122, it is determined whether there are corresponding connected components in the tree at nodes of greater depth than the current node. If this is the case, go to step 124, in which the current connected component is changed to the identified connected component and the current node to the corresponding node, and then to step 116. Conversely, if it is determined in step 122 that there is no corresponding connected component in the tree at nodes of greater depth than the current node, proceed to step 124. At step 124, it is scanned the set of related components of the graphic code and builds a corresponding tree. As explained with reference to FIG. 2, the preceding steps represent only one of the methods for scanning the graphic code and other strategies are possible for constructing the tree by associating with a given connected component of the graphic code a node of a tree whose number of sons is a function of the number of connected components contained in the given connected component.

In step 126, transforming the tree obtained in the previous steps into information. The process in this step is the opposite of the method described above with reference to step 14 of FIG. 2. For example, if step 14 is carried out using the procedure U described above, this will be the procedure F described in the same place that can be used in step 126. From an algorithmic point of view, the coding of Figure 3 as the decoding of Figure 6 can be implemented with limited resources. As regards the coding, step 14 can be performed, if one chooses to use the procedure U, in linear time; if the programming language used does not have automatic management of the memory or practical representation of the terms, it will be noticed that the number of nodes of the obtained tree is in O (m log m), and one will be able to work in a table fixed size.

Then comes a step of sizing and positioning the graphical representations of the nodes of the tree. The room for maneuvering this step is wide because of the commutativity of the tree and the deformability of the process. A simple and recursive way to realize this step is to represent each component by a rectangle with horizontal and vertical edges, and to reserve inside this rectangle one or two sub-rectangles to accommodate the recursively calculated representations of the nodes. This division can be done vertically or horizontally, this choice affecting the final size of the symbol if it is desired to maintain constant the distance between the edges of adjacent rectangles. A The algorithm can search for various configurations to minimize the total area, the largest dimension, or to give the symbol a desired aspect ratio.

As far as decoding is concerned, the algorithm supposes that it is possible to recognize in a digitized image all the connected components as well as their nesting relations. This can be done by scanning the pixels and using a Union-Find algorithm to update the sets of members of the components encountered as and when, which takes a memory amount of 0 {n) words machine and a time in 0 {n log ri) where n is the number of pixels, assuming that the memory is addressable in constant time. The Union-Find algorithms are well known and described in Introduction to Algorithms (Second Edition), Thomas H. Cormen, Charles E. Leiserson, Ronald L. Rivest, and Cliff Stein, MIT Press and McGraw-Hill.

The inventors have implemented the present invention using a program written in OCam1 language, compiled in native code for an x86 architecture machine; this machine received images either in real time by a black and white CCD sensor digitized by a Bt848-based acquisition card, with a resolution of 384 x 292, or by a digital camera with a resolution of 1600 x 1200 pixels . A graphic code of the type of that of Figure 4 or Figure 5, is decoded in less than 130ms in 384 x 292 (respectively 2.5s in 1600 x 1200) on a Xeon clocked at 3.00GHz. The calculation time is not very sensitive to the number of symbols present (with 125 copies of the symbols on an image 1600 x 1200, the time goes up to 2.75s). It depends essentially on the number of related components. Proper filtering of background noise can greatly reduce the number of spurious components.

FIG. 7 shows a graphic code coding the same information as the bar code of FIG. 1. In the example of FIG. 7, alternating black and white connected components were used, as in the example of FIG.

Figures 8 to 12 show different examples of graphic codes for information 777. Figure 8 shows the undeformed graphic code. Figures 9, 10 and 11 show different deformations of the code. These deformations could result either from a deformation of a physical medium on which the code is applied. They could result from the reading device used. These deformations do not modify the connectivity by arcs of the different components composing the graphic code and therefore do not modify the value of the coded information. The graphic code is thus resistant to any deformation that does not modify the connectivity of the components; it can in particular be decoded independently of axial stretching. FIG. 12 shows another graphic code for the number 777. In the example of FIG. 12 as in that of FIG. 5, connected components which do not have a specific and identical shape are used.

Figures 13 to 15 show another embodiment of the invention, in which a non-commutative tree is used. This means that the order of the threads in the tree is important. The graphic code is then oriented and the reading direction of the related components is important.

Figure 13 shows the tree corresponding to the number 12 - in binary 1100. The coding is as follows: a node with a son codes for a 1 while a node with 2 sons codes for a zero. Figure 14 shows a possible example of a graphic code, using a noncommutative binary tree. The graphic code uses rectangular connected components; it is oriented to account for the noncommutative nature of the tree. In the example, the threads of a node are ordered by their relative positions; for nodes of the same rank, the left son comes before the right son and the son of the top before the son of the bottom. The code of FIG. 14 corresponds to the information item 12, encoded according to the formula T (w) proposed above, with w = 1100. The reference numbers show the reading order. The mirror image of the mark of FIG. 14 does not therefore code the same information.

For a noncommutative tree, the decoding therefore implies, when searching for related components, to scan the tree in a given direction. The code remains resistant to uniaxial deformations; deformations can, however, lead to difficulties if the reading order is important. The advantage of this code is to present a higher density of information, at the cost of greater difficulty if it is desired to marry the code form recognizable by a human, similar to that of Figure 5.

Figure 15 shows another example of graphical code. In the example of Figure 15, the order of the connected components is given by the angle of the edges with the horizontal or the vertical. In other words, when a connected component contains two connected components, such as the component 1 and its two wires 2 and 3, it is first necessary to consider the connected component whose edges are parallel to the axes. Of course, it is only a reading convention, and one could use other conventions to differentiate two neighboring connected components, such as the thickness of the lines or more or less dark colors. Compared with the example of FIG. 14, the example of FIG. 15 only involves locally comparing two connected components and does not require orienting the graphic code.

The code of FIGS. 13 to 15 is resistant to deformations, like that of FIGS. 2 to 12. The explanations given with reference to Figures 2 and 6 are provided solely to facilitate the understanding of the invention. In practice, it is possible to use a graphical code transformation algorithm into a tree, which operates in a single pass and which makes it possible to directly find the tree corresponding to the graphic code, such as the one proposed above.

The graphic code of the invention, of which embodiments are shown in FIGS. 4, 5, 7 to 12, 14 and 15, is advantageously combined with the coded value, as represented in FIGS. 7 and 8. The coded value can be written in human readable form, for example under the graphic code. This allows on the one hand a reading of the combination by an automatic device and on the other hand a reading of the combination by a human operator. The graphic code may also simply be applied to any support, to allow recognition by a device of the type described with reference to FIG. 16. The support may be a label, a plastic film, or even a product itself. . The graphic codes of the invention can be characterized by various parameters:

- the depth of the tree associated with the code;

the number of nodes of the tree associated with the code;

the balanced nature of the tree, represented for example by the ratio between the number of nodes and an exponential function of its height; - the area occupied by the graphic mark, given the thickness of the line;

the information density, that is to say the ratio between the amount of coded information and the ratio of the area of the graphic mark and the square of the line thickness.

The depth of the tree is typically greater than or equal to five for the coding of usual information.

The number of nodes is typically greater than 10 for the coding of usual information.

The balanced character of the tree is appreciated by the ratio between the number of nodes and an exponential function of the height of the tree. For example, for a binary tree, consider the ratio N / 2 ^h , with N the number of nodes and h the height of the tree. This ratio tends to 1 for a complete binary tree; it tends to zero for a set of concentric circles. Coding with the methods described above provides values asymptotically greater than 0.1. For a p-ary tree, the number 2 is replaced by the number p in this formula; we find the number p by considering the maximum arity of the nodes of the tree, that is to say the maximum number of threads of a given node. Can also consider instead of the numeral 2 the value max (2, p), which also deals with particular cases, such as concentric circles. A code could then be characterized by the following inequality:

N max (2, /?) ^{Λ ~} with the definitions above. These three characteristics are representative of the tree used in the coding or the decoding. The following characteristics relate to the graphic mark and are representative of the fact that in the majority of applications a coding as compact or dense as possible is sought - which of course is not the case in FIG.

The area occupied by the graphic mark, given the thickness of the line, can be evaluated by the diameter of the largest disk of a homogeneous color or by the distance between adjacent connected components. This diameter or this distance is, in the examples of FIGS. 4, 7 and 8, of the order of the thickness of the line; this amounts to avoiding gaps in the mark. For example, the diameter or the distance may be less than or equal to four or even two times the line thickness used to write the mark. The density of information is assessed as the ratio between the amount of coded information, typically measured in bits, and the ratio of the area of the graphical mark to the square of the line weight. In the example of FIG. 4, four-bit information is coded in a substantially square graphic mark with a side of about 14 line thicknesses. The density is 1/50. In Figure 7, the mark has a size of about 80 x 40 line thicknesses, while the coded information is 40 bits long; this leads to a density of the order of 1.25%. Typically, the density of a graphic mark is greater than 0.5%.

To characterize a code, one of these criteria can be used, or a combination of any number of these criteria.

The invention also relates in another embodiment, a decoder apparatus. An example of such an apparatus is shown in the schematic view of FIG. 16. In the example of FIG. 16 is shown a support 160, with two graphic codes according to the invention. On the support 160 is also shown a reference; this mark is symbolic of the calculation of the position of the two graphic codes, as explained below.

The apparatus comprises an image acquisition part of the graphic code, of the kind described with reference to step 112 above. This acquisition part provides an image of the medium 160 and the graphic codes to a processing part 163. This acquisition part typically comprises an image sensor 161, followed by an analog / digital converter 162; of course, the converter can be integrated with the image sensor. The processing part 163 is adapted to decode the received image and outputs, for example to a display 164 or to a port the value of the information coded in the image. The portion 163 may comprise a conventional microprocessor, a programmable circuit or other logic means adapted to the automatic implementation of the decoding method. It is also possible to use wired logic, as indicated in FIG. 16 by the mention of a software or wired decoder. If necessary, one or more storage memories can be provided.

In the example of FIG. 16, the support 160 comprises two codes. The processing portion 163 is adapted to recognize the presence of both codes; in addition, it is adapted to provide an indication of the position of the two codes; in the example, this position is indicated relative to the Cartesian coordinate system represented in the figure. The reference 164 thus shows the display of the values of the two codes as well as the positions of these codes in the chosen reference. In the example, the landmark is a Cartesian landmark, given the rectangular shape of the acquired image; other types of markers may be used depending on the acquisition part. The invention is not limited to the examples proposed above. Thus, in the examples, only two colors were used for the related components. We could also use more than two colors, which increases the information at each level of the tree. For example, with three colors (white, gray and black), there are two choices for the color of each component, which allows to encode an additional bit per node. One could also use a single color for coding, meaning to attach meaning only to the related components of that color; this has the disadvantage of increasing the size of the mark, without making it more easily decipherable.

In the examples proposed above, binary trees were used; in the tree, each node has zero, one or two sons; each connected component therefore contains zero, one or two related components. One could also use more complex trees, for example ternary trees or (N-areas) in which the nodes can present up to three son (or N son). In this case, the related components of the graphic code could contain up to three (or N threads). This solution makes it possible to increase the compactness of the graphic code, without making the decoding more complex. The word "trace" has been used above to indicate the creation of a related component in the mark; the plot may consist of a physical writing of the connected component on a support; in this case, the tracing action is not necessarily implemented at each iteration of step 18, since it is possible to dispense with "tracing" white connected components. We can also "draw" in a memory the different related components, then proceed to a physical writing of the entire brand. In this case, the action of "plotting" amounts to assigning values to pixels of an image.

For this writing of the mark, it is possible to use all the known technologies, depending on the medium used for the graphic code and on the reading device envisaged. For reading, one can also use all available technologies; the term "color" must therefore be understood with reference to the technology used for reading; for example, an ultraviolet reading technology uses a "color", even though the ultraviolet is not strictly speaking a color. A tactile "read" technology would use as "color" differences in surface condition. As in the codes of the state of the art, error-correcting codes and redundancy can be provided to make the graphic coding more resistant. It is also possible to facilitate the acquisition of the code by reference marks, especially in the case of an orientation of the graphic code.

The coding of a word has been described with reference to the figures. Different words can be juxtaposed to encode information comprising several words.

Claims

A method of encoding information in the form of a graphical mark having a plurality of related components, comprising: a step of transforming the information to be encoded into a tree;

a step of transforming the tree into a graphic mark, in which a node of the tree is encoded by a connected component of the graphic mark comprising as many internal connected components as the present node of son.

2. The method of claim 1, characterized in that the tree is a commutative tree.

3. The method of claim 1, characterized in that the tree is a non-commutative tree.

The method of claim 1, 2 or 3, wherein the tree is a binary tree.

The method of one of claims 1 to 4, wherein the connected components have at least two colors, and wherein a node of the tree is encoded by a connected component of the graphic mark having a first color and comprising as many internal connected components of a second color as the node has son.

A method of decoding a graphical mark having a plurality of related components into an information, comprising:

a step of transforming the graphic mark into a tree, in which a node of the tree corresponds to a connected component of the graphic mark the node having as many threads as the corresponding connected component has internal connected components;

a step of transforming the tree into information.

7. The method of claim 6, characterized in that the tree is a commutative tree.

8. The method of claim 6, characterized in that the tree is a non-commutative tree.

The method of claim 6, 7 or 8, wherein the tree is a binary tree.

The method of one of claims 6 to 9, wherein the connected components have at least two colors, and wherein a node of the tree corresponding to a connected component of the graphic mark having a first color has as many son that the related component comprises internal connected components of a second color.

11. A support having a graphic mark capable of being decoded according to the method of one of claims 6 to 10.

12. A support having a graphic mark obtained by the coding method of one of claims 1 to 5.

13. The medium of claim 11 or 12, further having an indication of the value of the information decoded or encoded according to said method.

14. The support of the combination of claim 11, 12 or 13, wherein a connected component of the graphic mark has a rectangular shape.

15. The medium of one of claims 11 to 14, wherein the tree corresponding to the mark in the coding or decoding has a height greater than or equal to 5.

16. The medium of one of claims 11 to 15, wherein the tree corresponding to the mark in the coding or decoding has a number of nodes greater than 10.

17. The medium of one of claims 11 to 16, wherein the tree corresponding to the mark in the encoding or decoding satisfies! , _Λ ≥0,1 max \ 2, p) where N is the number of nodes, p is the maximum arity of the nodes of the tree and h is the height of the tree.

18. The support of one of claims 11 to 17, wherein the diameter of the largest homogeneous color disc in the mark is less than or equal to four times the line thickness used to write the mark.

19. The medium of one of claims 11 to 18, wherein the density, expressed by the ratio between

the quantity of coded information, measured in bits and

the ratio of the area of the graphic mark and the square of the line thickness is greater than 0.5%.

20. A decoder apparatus, comprising:

an acquisition part (161, 162) adapted to provide an image of a graphic code;

a processing part (162, 163) that can be implemented by software or cable and adapted to receive the image supplied by the acquisition part and to decode the received image (160) to provide the value of the coded information in the graphic code.

The apparatus of claim 20, wherein the processing portion is adapted to provide an indication of the position of a graphic code in the received image.

22. A computer program on a computer readable medium containing code adapted to execute on a computer all the steps of the coding method of one of claims 1 to 5.

23. A computer program on a computer-readable medium containing code suitable for executing on a computer all the steps of the decoding method of one of claims 6 to 10.