WO2020144723A1

WO2020144723A1 - System for calculating curvature values at each point of a road route and operation method thereof

Info

Publication number: WO2020144723A1
Application number: PCT/IT2020/050002
Authority: WO
Inventors: Paolo ANDREUCCI; Anna ANDREUSSI; Cristina CAMARDELLA; Antonio Frisoli; Claudio LOCONSOLE
Original assignee: 9piu' Srl
Current assignee: 9piu' Srl
Priority date: 2019-01-09
Filing date: 2020-01-09
Publication date: 2020-07-16
Anticipated expiration: 2021-07-09
Also published as: EP3908806A1; IT201900000220A1

Abstract

Described is a system (1) for calculating curvature values (V_ci) of a curve (C_i) in a road route comprising storage means (21) for storing at least one satellite cartographic map containing said road route, a logic control unit (2), operatively connected to said storage means (21) for retrieving said at least one satellite cartographic map, said logic control unit (2) being equipped with processing means (22, 23) for processing said at least one satellite cartographic map, comprising a calculation program configured for calculating the curvature values (V_ci) in each point of said road route, and for associating to said curvature values (V_ci) a data structure whose distribution is variable corresponding to said calculated curvature values (V_ci). The invention also relates to method for operation of said system.

Description

System for calculating curvature values at each point of a road route and operation method thereof.

This invention relates to a system for calculating the curvature values at each point of a road route.

The invention also relates to the operating method of the system.

More in detail, the invention relates to a system of the said type, designed and realised in particular for calculating the curvature value of one or more curves present along a road route, starting from one or more cartographic maps, in such a way that a driver can adopt a suitable behaviour for executing the curve in a safe and high-performance manner, but which can be used for any purpose, for which it is necessary to know in advance the curvature value of a curve present along a road route.

The description below relates to a system for calculating curvature values present along a road route and the operation method thereof, but it is quite apparent how the same should not be considered limited to this specific use.

As is well known, systems and methods currently exist for detection and graphical representation of road routes which allow a driver to know in advance the morphology of a road route to be travelled along.

Some of these known systems are based on mathematical methods for calculating the value of the curvature based on the identification of three points of a curve and tracing ideal straight lines which join together each pair of points which are formed with the three points.

Mathematical processing is then performed, such as interpolations, to obtain median reference points.

It is evident that this procedure is onerous in terms of mathematical calculation correlated with the process and is subject to approximation errors such that it is not very precise or reliable.

There are also prior art systems which process data coming from cartographic maps in order to obtain the angle of intersection between two straight lines which represent a road entering into a curve and a road exiting from a curve, in order to roughly determine the direction of the curve and its curvature value.

These systems also comprise a rather approximate processing of cartographic data which does not provide reliable and precise details, especially for users who want to drive safely or want to obtain certain levels of performance during sports driving, by knowing in advance the trend of the road route.

In light of the above, the aim of the invention is therefore to provide a system for calculating the curvature value of a curve based on a simple and reliable processing.

A further aim of this invention is to provide the tools necessary for the execution of the method and the apparatus which execute the method.

The specific object of this invention is therefore a system for calculating curvature values of a curve in a road route comprising storage means for storing at least one satellite cartographic map containing said road route, a logic control unit, operatively connected to said storage means for retrieving said at least one satellite cartographic map, said logic control unit being equipped with means for processing said at least one satellite cartographic map, comprising a calculation program configured for calculating the curvature values in each point of said road route, and for associating to said curvature values a data structure whose distribution is variable corresponding to said calculated curvature values.

Further according to the invention, said data structure is a polyline comprising a plurality of points having a variable density so that, in correspondence with curvature values that exceed a predetermined threshold, the points of said polyline have a density higher than the density of the points of the polyline corresponding to curvature values lower than said predetermined threshold.

Again according to the invention, said system can comprise a driving unit connected to said logic control unit, and which can be connected to the operating control unit of the vehicle, on which said system can be installed, configured for receiving as input and processing said calculated curvature values, the physical parameters of the vehicle, the forward speed value of said vehicle and geographical positioning coordinates of the vehicle, and for sending corresponding control signals to said control unit, so as to brake or accelerate the forward speed of the vehicle.

Preferably, according to the invention, said system comprises a display unit for displaying said road route on said satellite cartographic map, with the data structure associated with it.

A further objection of the invention is an operation method of a system for calculating curvature values of a curve in a road route, comprising the following steps:

a. generating an image associated to a predetermined portion of a satellite cartographic map, starting from a retrieved satellite cartographic map, containing a road route;

b. if said image has not a predetermined resolution, generating a NxN matrix of sub-images;

c. if the step b has been carried out, blending said sub-images of said NxN matrix, to obtain a further image, different from said image;

d. removing one or more indicators, such as labels or points of interest present on said retrieved satellite cartographic map, from said image, or from said further image;

e. converting said image, or said further image into a binary image; f. generating a skeletonized image starting from said binary image; g. detecting and removing one or more further road routes with respect to said road route of interest, present in said skeletonized image; h. generating spatial coordinates for each point of said road route; i. analysing the gradients in each of said points;

j. classifying the curves present in said road route;

k. generating a polyline having a variable density.

Further according to the invention, said binary image generated in said step f. comprises a white pattern comprising a plurality of pixels in which each pixel uniquely identifies a point of the center of the road of said road route, and a pair of geodesic coordinates of the point itself is associated to each pixel.

Again according to the invention, in said steps h. and i. said processing unit calculates the successive derivatives on the set of said geodesic coordinates extracted in said step f. to obtain the curvature value in each point of said road route.

Preferably, according to the invention, in said step j. thresholds of said curvature values calculated in said steps h. and i. are established and difficulty levels are associated with said thresholds.

Further according to the invention, in said step k., said processing unit receives as input the set of said geodetic coordinates belonging to said road route, the classification data of said step j. and it performs a variable sampling of the curves present in said road route, in particular it makes a selection of a sub-set of coordinates of points belonging to a determined curve.

The invention is now described, by way of example and without limiting the scope of the invention, with reference to the accompanying drawings which illustrate preferred embodiments of it, in which:

Figure 1 shows a block diagram of the system for calculating the curvature value of a curve according to this invention;

Figure 2 shows a schematic image of a 3X3 matrix associated with a satellite cartographic map;

Figure 3 shows a schematic image of the matrix of Figure 2 processed;

Figure 4 shows an image SF which is the result of a processing step of the method of operation of the system according to the invention, in particular of the removal of labels and points of interest present on the cartographic map;

Figure 5 shows an image SE which is the result of a processing step of the method of operation of the system according to the invention, in particular of the blending of the images of the matric of Figures 2 and 3;

Figure 6 shows an image SB which is the result of a processing step of the method of operation of the system according to the invention, in particular of the extraction of a binary image;

Figure 7 shows a schematic image SK which is the result of a processing step of the method of operation of the system according to the invention, in particular of the skeletonization operation;

Figure 8 shows a schematic image of the result of a processing step, in particular of the identification of multiple routes of the image;

Figure 9 shows a graph obtained as the result of a processing step of the method of operation of the system according to the invention, in particular of the extraction of Cartesian coordinates;

Figure 10A shows a graph which contains the values of the X-axis of the curve of the graph of Figure 9;

Figure 10B shows a graph which contains the values of the Y-axis of the curve of the graph of Figure 9;

Figure 1 1 shows an schematic image of the result of a processing step, in particular the analysis of the gradients of the curve with three levels of difficulty, easy, medium and difficult;

Figure 12 shows an schematic image of the result of a processing step, in particular a labelled map with a basic level of detail; and

Figure 13 shows a schematic image of the result of a processing step, in particular a polyline with different resolution in the points of greatest difficulty.

The similar parts will be indicated in the various drawings with the same numerical references.

The system 1 for calculating curvature values of a curve, according to the invention, can be installed on a vehicle such as a car, a sports car or motorcycle and the like.

Said system 1 for calculating curvature values of a curve basically comprises:

- a logic control unit 2 equipped with a unit for data retrieval, processing and storage which is described in detail below;

- a user interface unit 3, connected to said logic control unit 2, by means of which the user can enter data in said system 1 ;

- a display unit 4, on which a user can display the result of the data processing; and

- a driving unit 5, connected to said logic control unit 2, which allows a control signal to be sent to a control unit of the vehicle, as a function of the degree of difficulty of the curve, such as, for example, the speed limit, the variation of the driving angle, the acceleration and the like.

Said logic control unit 2 comprises inside it a storage means 21 in which are stored one or more known satellite cartographic maps.

Said storage means 21 can also be positioned outside said logic control unit 2, without thereby departing from the scope of the invention.

Moreover, said logic control unit 2 can also be connected by cable or via wireless or Bluetooth mode with a remote unit outside said system 1 from which to retrieve cartographic maps of interest.

Said logic control unit 2 comprises inside it a retrieving unit 22 connected to said storage means 21 for retrieving one or more satellite cartographic maps.

Alternatively, said retrieving unit 22 can be outside said logic control unit 2.

Moreover, said retrieving unit 22 can be connected with said external remote unit.

Said logic control unit 2 also comprises a processing unit 23 designed for the processing of input data coming from said storage means 21 and retrieving unit 22, by means of a calculation program.

Alternatively, said processing unit 23 can be outside said logic control unit 2.

Said user interface unit 3 allows the user to set up the data in said system 1.

For this reason, said interface unit 3 can be connected to said logic control unit 2.

Said user interface unit 3 is an input unit, such as a web page, a desktop application, a smartphone application or a stand-alone application on a custom device with dedicated hardware and the like.

Said display unit allows the user to display the output data from said system 1 which are supplied from said logic control unit 2 and/or from said interface unit 3.

In particular, the data structure, that is to say, the polyline, comprises a plurality of points which are the geographical coordinates of the polyline.

The geographical coordinate are used, by means of a special mapping between geographical and spatial coordinate of the image, to indicate the points of the polyline on the map itself.

These points are then connected to show on the display unit 4 the route to navigate, that is to say, the road route which the user must follow.

Said display unit 4 comprises a display.

Said display can be the display of a smartphone, a monitor, a projector or a head display, that is, an information projection device on the front windscreen of a car or of a vehicle in general.

Said driving unit 5 is an accessory unit which receives as input the values of the curves, their classification, the physical parameters of the vehicle on which said system 1 is installed, the forward speed value of said vehicle and the geographical positioning coordinates of the vehicle.

Said driving unit 5, on the basis of the data received as input, sends control signals to the control unit of the vehicle, so as to vary the forward speed of the vehicle, braking or accelerating the speed of the vehicle.

Alternatively, in the absence of said driving unit 5, the driver of the vehicle can display the route on the display unit 4 and adapt the driving on the basis of the classification of the curves performed by the system 1.

In particular, said display unit 4 displays the road route with the graphical representation of the data structure associated with the route and with the instantaneous localisation of the vehicle on the route, in such a way that the driver can accelerate or brake the forward speed of the vehicle as a function of its current position on the route and the type of curve to be travelled along.

The operation of said system 1 described above is performed in the following steps:

a. generating an image associated to a predetermined portion of a satellite cartographic map, starting from a retrieved satellite cartographic map;

b. if necessary, generating a NxN matrix of sub-images SN,N;

c. if the step b has been carried out, blending said sub-images SN,N of said NxN matrix, to obtain a further image SF, different from said image S;

d. removing one or more indicators from said image S, or from said further image SF, such as, for example, labels or points of interest present on the retrieved satellite cartographic map, obtaining an image SE;

e. converting said image SE, or said further image SF into a binary image SB;

f. generating a skeletonized image SK starting from said binary image SB;

g. detecting and removing one or more further road routes with respect to said road route of interest, present in said skeletonized image Sk;

h. generating spatial coordinates;

/. analysing gradients;

j. classifying the curves;

k. generating a polyline or data structure with a variable density, that is to say, the distribution of the data or elements contained in the data structure is variable.

When a user intends to follow a road route, the user activates said system 1 and enters in said interface unit 3 an address of a geographical point which is to be geolocated.

Said geographical point is identified by means of geographical coordinates such as, for example, latitudes and longitudes.

Said retrieving unit 22 retrieves the satellite cartographic map corresponding to an area positioned about said geographical point.

Said satellite cartographic map can be stored in the storage means 21 or in the remote external unit.

Once the cartographic map of interest has been retrieved, said system 1 processes said satellite cartographic map so as to extract an image S relative to said area, according to the scale index and extension settings of the image set by the user by means of said interface unit 3.

Said area is consequently an area of predetermined dimensions selected by the user. Said satellite cartographic map retrieve is processed as if it were a planar map, since both the curvature of the terrestrial surface and the altitude contribute in a negligible manner to the determination of the curvature value of a curve.

Said processing unit 23 implements, by means of said calculation program, an algorithm which provides as output a calculation of the curvature values V_d in each point of said road route, and the association to said curvature values Vd of a data structure whose distribution is variable corresponding to said calculated curvature values Vd, as described in detail below.

In particular, in the step a., the processing unit 23 receives said satellite cartographic map from said retrieving unit 22 and associates with it the following metadata: coordinates of the centre of the S, also called barycentre or centroid, the scale index and the extension or the dimension of the image S measured in pixels or in metres.

The latitude and longitude geographical coordinates of the geographical point are the centroid or barycentre of the image S.

The scale index of the image S to be obtained is also called zoom.

This is understood to be a parameter which allows the dimension in pixels of the image to be obtained and the physical dimensions of the starting satellite cartographic map to be put in relation.

The dimension and the scale index of the image S allows a fixed area of extension which surrounds the geo-localized point to be uniquely identified.

If the image S, which includes the entire route or the point to be geolocated, does not have the desired resolution and dimensions, that is to say, if the original image does not possess the correct resolution, due to the fact that some satellite services, from which the satellite cartographic map is obtained, allow an image to be generated with a maximum dimension in pixels, and consequently provide a certain resolution, it is possible that the entire image does not cover the entire zone to be geolocated, it will be necessary to generate sub-images SN.N, according to step b., arranged in a NxN matrix of images, all of the same dimension, at the desired resolution and subsequently blend them together in an image SF, which will have the desired characteristics.

With reference to Figure 2, the step b. is performed as follows: using the above-mentioned metadata, said processing unit 23 generates an NxN matrix of square sub-images SN.N, all of the same dimension and sequential, M metres from each another, where M is the distance between the respective centroids, and where N is the smallest whole number which allows, in the step for blending sub-images SN.N, the generation of a further image SF or reconstruction of the image S which originally included the entire route or point to be geolocated.

For this reason, N can be defined as the rounding up of the ratio between the side, in metres, of the square image S and the side, in metres, of the generic sub-image SN,N which is also square:

where

ds is the dimension, in metres, of a side of the image S;

dsN,N is the dimension, in metres, of a side of a sub-image SN,N.

The centroid provided initially is the centroid of the sub-image Si,i of the NxN matrix.

The dimension of the sub-image SN,N is selected, for example, equal to 640x640 pixels or a size which enables a sub-image SN,N to be obtained with both sides equal.

The resolution of the generic sub-image SN,N may be calculated with the following ratio:

U = ft (b)

where I indicates the value in pixels of the side of the generic sub image SN,N.

An index o, expressed as a percentage, for the superposing of two adjacent sub-images is set, both in the horizontal and vertical direction of the matrix, that is to say, between the image SN,N and, for example, the image SN, N_{+I ,} said index o being included in a range from 0%, corresponding to two sub-images being perfectly spatially adjacent, to 50%, corresponding to the configuration according to which the centroid of the next sub-image SN, N_+I lies on the edge of the previous sub-image SN,N.

Typically, the superposing index o is selected with the smallest possible value so as to reduce the number of sub-images SN,N which cover the extension of the starting map but which, however, guarantees a subsequent blending of the correct sub-images.

For this reason, generically, with a superposing index o close to 0, the number of pixels in common between one sub-image and the adjacent one might not be sufficient to allow the blending algorithm to be completed correctly.

Typically, a superposing index o of about 10% ensures the correct performance of the blending.

Moreover, by setting a superposing index o different from 0, it is necessary to perform a modification to the number N of sub-images which make up the NxN matrix, modifying the formula (a) as follows:

The distance in degrees is then calculated between the centroids of the two adjacent sub-images SN,N and SN, N_+I positioned, for example, along a first line of the NxN matrix, according to the following formula: dlat,6 = [M-(o*l)*U]/[R(lat) ] (d)

άΐoh,q = [M-(S*l)*U]/[R(lat)s n(n/2-lat)] (e) where / indicates the value in pixels of the side of the generic sub image SN.N, or the superposing index of the sub-images, lat the value of the latitude selected and R {lat) is the value of the terrestrial radius relative to the latitude selected, which is calculated as follows:

sin(lat))² (f) where n corresponds to the radius at the equator and r2 corresponds to the radius at the poles (in the latitude direction).

Lastly, diat,e is added to the current value of latitude, if the lower sub-image is being calculated, or diat,e is subtracted from the current value of latitude, if the upper sub-image is being calculated, or di₀ is added to the current value of longitude, if the right sub-image is being calculated, or dion,e is subtracted from the current value of longitude, if the left sub-image is being calculated.

By way of example, starting from the sub-image S2,2 having coordinates Iats2,2 and lons2,2, for generating the lower sub-image S3, 2 the following calculation is performed: Iats3,2 = Iats2,2 - dia C,Q, whilst lons3,2 = lons2,2.

For generating, on the other hand, the upper sub-image S-1,2, the following calculation is performed: lats-1,2 = Iats2,2 + dia C,Q, whilst lons-1,2 = lons2,2.

For generating, on the other hand, the right sub-image S2,3, the following calculation is performed: Iats2,3 = Iats2,2, whilst lons2,3 = lons2,2 + dlor,e.

For generating, on the other hand, the left sub-image S2,i, the following calculation is performed: Iats2,i = Iats2,2, whilst lons2,i = lons2,2 - dlor,e.

In this way, by means of the values diat,e and di_on,e it is possible to obtain a shift of the centroid of the generic sub-image SN,N of approximately M metres less the superposing value expressed in metres.

Iterations of this process are performed to update the current values of latitude and longitude of each centroid of each sub-image SN.N, up to the completion of the NxN matrix.

In the calculation of the shift in degrees in the latitude direction, for generating upper or lower sub-images, the variation of the terrestrial radius is considered to be negligible, and is therefore considered to be constant; the metres/pixel resolution inside the sub-image is considered to be constant, since the variation of the terrestrial radius using the selected sale index is considered to be negligible.

The subsequent steps of the processing method are performed either on the initial image S, or, if it has been necessary to generate the NxN matrix of sub-images SN.N, it is firstly necessary to blend/superpose the sub-images SN,N in order to reconstruct a further image SF and then perform the further steps on the further image SF.

For this reason, in order to blend/superpose the sub-images SN,N the following step c. is performed by means of said processing unit 23.

The n-th row of the NxN matrix, for example row 1 , is selected.

A pair of adjacent sub-images is selected, for example the pair Si,i and Si ,2.

It should be noted that each sub-image SN,N is a matrix of pixels with / rows and / columns.

Both the sub-images Si,i and Si,2 are divided in half in the height direction, that is to say, in a vertical direction, thereby obtaining, with reference to Figure 3:

the first half S’i,i of the sub-image Si,i and the second half S”i,i of the sub-image Si,i

the first half S’1,2 of the sub-image Si,2 and the second half S”i,2 of the sub-image S-1,2.

The second half S”i,i of the sub-image Si,i and the first half S’1,2 of the sub-image Si,2 is selected, respectively, a counter / is set to 0, a variable in the form of vector V with an array of /si,i/2 elements equal to zero, which will count the correlation values between the /-th sections of the sub-images, a variable h, that is, the superposing variable, at 0 and an iteration is performed, on the sub-images selected, of the following algorithm, until the variable p reaches the halved value /si,i of the width of the sub-image Si,i less 1 .

The algorithm is as follows: / columns of pixels are removed from the start of the first half S’1,1 and from the end of the second half S”i,i of the sub-image Si,i .

The correlation is calculated between the section consisting of the / columns removed from the second half S”i,i of the sub-image Si,i and the section consisting of the / removed from the first half S’1,2 of the sub-image Si,2, and the value is saved in the vector V, at position /.

Upon completion of the iterations, that is, when the variable V has reached the value /si,i/2, the maximum index of that vector is found and it is assigned to the superposing variable h. h columns of pixels corresponding to the value of the superposing variable h are removed from the sub-image Si,i starting from the left, and this sub-image

is blended, modified with the sub-image Si taken in its entirety.

If the number of columns of the NxN matrix has been reached, the result of the blending of the sub-images SI ,N of the row 1 is saved, expressed as matrix lxl₀ of pixels, where lo is the number of columns which depends on how the sub-images in the previous algorithm have been superposed, so l₀<N^*l on the basis of that said on the percentage of superposing o, and a new line of the NxN matrix is selected, otherwise a new pair of adjacent sub-images is selected, taking as the first sub-image the one just generated and as the second sub-image the one immediately to the right in the NxN matrix.

If the number of lines of the NxN matrix has been reached, the iteration is ended on the lines of the algorithm.

At the end of this step, N images corresponding to the blending of the sub-images belonging to the N rows of the NxN matrix are generated, that is to say, N matrixes of lxl₀ pixels which constitute the result of the N blends saved previously, so a Nx1 matrix is obtained.

After having generated the N blended images corresponding to the N lines, in order to obtain the global image SF the algorithm is re-launched on the latter, starting from the selection of the pair of blended sub-images, up to the end of the N rows.

By way of example, in a 2x3 matrix, in which there are, therefore, 6 sub-images, the first step of the algorithm generates 2 blended images which are the result of the blending of the 3 sub-images present in the first column.

After having obtained these 2 blended images, it is necessary to blend them in the vertical direction of the matrix, so the algorithm restarts to perform the blending along the 3 columns, thereby obtaining the complete blended image SF.

As shown in Figure 3, an image SF is obtained which is the result of the blending of the sub-images SN,N of the NxN matrix. As mentioned above, if the sub-images SN,N have not been generated and subsequently blended, the image SF coincides with the image S.

Figure 5 shows an image SE, obtained from the image SF after having carried out the step d., that is, removal of any labels, where the term“labels” means those indicators or symbols present on the original satellite cartographic maps, such as names of towns, districts, shops, restaurants, service stations, parks and the like, or details referred to scale indices and the like as they constitute a disturbing element for the subsequent operations to be performed on the image.

The step d. is performed according to prior art techniques, that is to say, cutting the image until completely excluding the section of the labels.

The points of interest, on the other hand, if present, can be located anywhere in the image and, therefore, the most convenient method for eliminating them is by entering predetermined parameters in the request of the image itself, through the remote web services of the provider of the cartographic maps used.

Step e. is then performed, generating from the image SE a binary image SB, as shown in Figure 6, in order to extract the coordinate relative to the position, in pixels, of the non-zero pints of the image.

In particular, the image SE is initially transformed into a grayscale.

Subsequently, a threshold value T of the intensity of pixels of the image SE IS selected.

The value of the threshold T is calculated, with known methods, by analysing the coloured histograms of the above-mentioned image SE.

All the pixels of the image SE are analysed; all the pixels with a value above the threshold value T are assigned with the value 255, whilst the rest are assigned the value 0.

In this way, the image SB is obtained in which the route which can be identified is shown in white, whilst all the rest is shown in black.

Step f. is then carried out, in which, with reference to Figure 7, starting from the binary image SB, a“skeletal” image SK is obtained by means of the skeletonization operation which reduces to a single pixel the thickness of the white route or of the white part of the binary image SB, starting from the middle part of the thickness.

In particular, each pixel of the white route uniquely identifies a point of the centre of the road of the route and, therefore, of the curves present in the route, so it is possible to immediately determine the geographical coordinates of the points of the route which are on the white route, linking each white pixel with a pair of coordinates, through the conversion parameters supplied by the provider of the cartographic maps.

These parameters link the coordinates in pixels of each point of an image, representing a cartographic map, to the geodetic coordinates of the same points in the relative reference system.

Moreover, the skeletonization operation also allows any isolated points of the black background or spurious points to be eliminated, resulting from false measurements of the road route on the satellite cartographic map of origin.

Step g is then performed, for detecting and removing one or more further road routes with respect to said road route of interest, present in the skeletonized image SK.

In fact, during the performance of the algorithm it is not possible to determine beforehand the quantity of road routes contiguous with the one of interest present inside a sub-image SN,N and any presence of crossings creates a difficulty in the measurement of the quantity of pixels associated with a curve.

The measurement of the road routes which are not of interest occurs in accordance with the following sub-steps performed by said processing unit 23, with reference also to Figure 8.

The presence of points of the white route along the edges of the skeletonized image SK is detected in sub-step g.1. and one of these is selected as the starting point. If the skeletonized image SK has not been generated, the operation is carried out in the same manner starting from one of the edges of the image S.

In step g.2. a small square area is selected, for example with a size of 3x3 pixels, inside of which to verify the contiguous nature of the route: this means verifying the presence of white pixels inside that area, with the exception of the starting point.

A vector C is initialised, called the vector of the crossings which will contain the coordinates (x, y) of the points of the crossings, if these are detected during the exploration of the first route.

A variable p corresponding to the number of the route which is being analysed is initialised.

In step g.3. the value of the current route, that is, of the variable p, is assigned to the starting point, for example first route: value 2, second route: value 3 and so on, and a new element is created in the list of routes stored in said storage means 21 as vector C which will contain the coordinates (x,y), in the skeletonized image SK, of the white pixels of the p-th route.

Step g.4. verifies whether there are more than 3 pixels with logic value 1 are present in the area; if so, a crossing has been detected and this point is inserted in the vector C.

Step g.5. detects and enters in the route the pixel of the white route (value in the logic matrix equal to 1 ) closest to the point of the route, where the vicinity is expressed as the sum of the distance along the height and the width of the image, between the current point of the route and each point of the white route of the square area defined previously.

If two points of the white route exist at the same distance, one is selected in a random fashion.

In step g.6., when one of the edges of the image S or sub-image SN,N is reached, the content of the vector of the crossings is evaluated: if it is empty then the algorithm finishes, otherwise the value of the current route is assigned to the starting point, the point is removed from the list of crossings and it is assigned to the starting point.

At the end of the algorithm it is possible to display n different routes and the relative spatial coordinates are stored in said storage means 21 or in a remote device, for the subsequent processing.

If the aim is to analyse one of the n routes in particular or a combination of them, then it is possible to supply a set of driving coordinates to the algorithm by means of which, by calculating the Euclidean distance between each point of the n-th route and each driving coordinate, it is possible to generate the final route since the distance between a driving coordinate and each point of the n-th route will be a minimum in the closest point, that is to say, the desired one; by iterating this process for all the driving coordinates, a route or a combination of them is automatically selected, selecting all the points of the route of interest between that in which the minimum i-th has been identified and that in which the minimum i+1 -th has been identified.

The final route is called ‘extracted route’ also if all the n-routes determined in step g.6 is to be analysed, without selecting one in particular.

Step h. is then carried out, with reference to Figure 9, to extract the Cartesian coordinates of each point present on the white route and subsequently to calculate the gradients of the curve and label them, that is to say, defining a level of difficulty.

For this purpose, said processing unit 23 must analyse the successive derivatives of the function associated with the route extracted and select inside it a discrete set of values Xn, the threshold values which separate the same number of levels of difficulty of travelling along the curve of the road route.

In particular, in step /., on the basis of the degree of precision to be assigned to the operation of labelling the curves, a number X_n of thresholds of levels of difficulty for travelling along the curve is defined: these thresholds are each identified by a curvature value Vci which is not dependent on the route being analysed and therefore determined in advance. These values are expressed as values of the second derivative, point by point inside the route to be labelled.

The number of thresholds Xn starts from a minimum value XMIN, corresponding to the minimum curvature value VMIN, for example relative to a straight line, and reaches a maximum value XMAX, corresponding to the maximum possible curvature value VMAX, for example relative to a‘IT curve. Therefore, using Ci to indicate a generic curve present in the route in question, this will be assigned with a predetermined level of difficulty or label Dei if its curvature value Vci exceeds a predetermined curvature threshold Xn, remaining, however, less than the curvature threshold immediately higher X_n+i.

In order to perform this assignment, use is made of the coordinates stored previously in step g.6, belonging to the route extracted, multiplying them by the metres/pixel resolution value calculated previously during the selection of the map: in this way, each pair of coordinates is expressed in pixels, in pairs of coordinates, in metres, which contain the distance value with respect to the point positioned at the top left end of the skeletonized image SK.

With reference to Figures 10A and 10B, by applying the second derivative on the series of coordinate in metres just calculated, it is possible to calculate the curvature value Vci at each point and subsequently assign the level of difficulty or label Dei.

Subsequently, step j. is performed for classification of the curves Ci on the basis of the level of difficulty Dei which is to be identified and, therefore, on the basis of the number of thresholds Xn set.

Figure 1 1 shows, by way of example, 3 levels of difficulty Dei associated with 3 different curvature thresholds Xs.

The levels of difficulty Dei are displayed by means of horizontal straight lines of different colour: each straight line delimits the start of the relative curve difficulty and all the curvature values Vci, on the Y-axis, between this straight line and that of the immediately subsequent difficulty, constitute the range of difficulty Dei considered.

The example indicates the following levels of difficulty:

1. easy (level Di);

2. average (level D2);

3. difficult (level D₃).

Without departing from the scope of the invention, it is also possible to fix a number of levels of difficulty different from three, on the basis of the level of precision to be obtained in the classification of the curves.

The X-axis shows the points of the extracted route which correspond to the pairs of coordinates, in metres, on the route, and the Y- axis shows the curvature value Vci at that point.

Considering the calculation of the second derivative, it is possible to obtain, depending on the morphology of the curve, values of curvature Va which are also negative, but for the purpose of the analysis of the above- mentioned derivative, the outcome of the labelling operation is identical if an absolute value is applied to the function which can be seen in Figure 11 on each single sample.

Obviously, the number of thresholds Xn which can be identified is variable and depends on the level of detail to be provided.

In general, for the labelling step

vectors Yi are created, called vectors of the curves, created with the aim of managing the curves Ci of the route in such a way as to have for each vector and, therefore, for each level of difficulty Dei, the curves belonging to that particular level of difficulty Da

The vectors Yi are vectors which contain binary values: 1 if the point of the route belongs to a curve with relative difficulty of that vector, and 0 elsewhere.

In each vector Yi the values are all equal to 0, and, where the curvature value on the j-th point of the route exceeds the threshold Xn the value of the j-th element of the vector Yi associated with that level of difficulty Dei adopts value 1 , otherwise value 0, that is, the one already present.

In that way it is possible to represent the labelling of the curves on the route, since, because each element of the vector Yi, which is associated with a precise level of difficulty Dei is associated with a point of the route extracted, this point of the rote will automatically adopt the assigned difficulty Da

With reference to Figure 12, the labelling or classification step of the curves is then performed, which is carried out by said processing unit 23 following the analysis of the gradients.

This information is displayed by means of said display unit 4 in text form.

A level of difficulty Dei, numerical or descriptive, is uniquely associated to a curve Ci or to a portion of it.

It is possible to label the curve Ci with different levels of detail: generic difficulty of the curve C\ entry and exit difficulty of the curve C\ entry, maintaining and exit difficulty of the curve Ci.

With these levels of difficulty Dei one can choose whether to represent on the map, or to save in an extracted structure, each single variation of the difficulty of the curve, exactly as described by the curve vectors Yi defined previously, or only the maximum of the difficulty of that curve.

By using the latitudinal and longitudinal coordinates of the central point of the image S, the scale index, the size of the original satellite cartographic map and the levels of difficulty Dei, it is possible to associate, in real time, a coordinate to a curve, if present, indicating with a note the relative difficulty of travel.

Moreover, it is possible to modify this difficulty as a function of the speed of the vehicle, a curve considered“Simple” becomes“Media” if the speed in turn exceeds a certain threshold.

With reference to Figure 13, step k. is then performed, with the creation of a polyline with a high resolution and variable density.

In particular, as mentioned above, the expression “polyline with variable density” means a data structure whose distribution is variable, in particular corresponding to the calculated curvature values Vd.

In more detail, the variability of the quantity of data or points contained in the polyline depends on the radius of curvature of the curve itself; for a straight stretch of the route there are fewer points with respect to a curve with a small radius of curvature. Said processing unit 23, starting from the set of coordinates belonging to the extracted route, selects, on the basis of the classification of curves created previously, to sample the curve in a more or less dense manner, that is to say, selecting only a sub-set of di coordinates of points belonging to that curve.

In order to integrate the information on the geolocated curves with various applications, it is possible to construct extracted data structures, for example classes.

A possibility is to represent the data through a polyline, which is an ordered and finished set of segments orientated in an orderly consecutive manner, that is contiguous, that is to say, the second end, in the direction of orientation, of a segment coincides with the first end of the next one, or a set of points, each point being a pair of coordinates of the extracted route.

The denser the polyline, that is to say, the greater the density of the points, the less will be the difference between the latter and the actual curve.

The resolution of the polyline according to the invention depends on the number of points which the curve has after the transformation into Cartesian coordinates.

The usefulness of a greater number of points on the road route means in the most dangerous points of the route: the more the curve is difficult the denser the polyline must be in order to improve the travelling instructions.

By using the curve vectors Yi mentioned above it is possible to construct a final polyline which possesses, for example, a fourth of the points in the sections with simple difficulty, a third of the points in the sections with medium difficulty and all the points extracted from the original image in the sections with high difficulty.

This choice depends solely on the system which will use the polyline: if it is a high performance system it is not necessary to carry out any sub-sampling, if it is a mobile system, such as a smartphone or a low- power device, then the reduced processing capacity represents a limit to the quantity of points which can be processed in real time.

The values Ki are whole values called“jumps” and they are as many as the levels of difficulty Dei and each Ki indicates the frequency of sampling of the route extracted which has a level of difficulty Dei. The values Ki are defined before the sub-sampling step of the extracted route and depend on the computational capacity of the support designed for the real-time processing of the data structure of the route.

For every element of the vector of the curves Yi, which is associated with a level of difficulty Dei, the value of which is equal to 1 , the pair of coordinate at the index of that point in the extracted route is extracted and Ki subsequent elements are skipped in the route extracted, again associated with the difficulty Dei.

By way of example, if the first 30 elements of the vector Yi are all equal to 1 and the vector Yi is associated with a level of difficulty Dei corresponding to the ‘Simple’ label, it means that the first 30 pairs of coordinates of the route extracted have been labelled as ‘Simple’, for example belonging to a not very sharp curve.

If a Ki equal to 3 has been defined, then, then after having selected the first pair of coordinates, corresponding to the first occurrence of T in the vector Yi, always at the index 1 , and having inserted it in the final polyline, the subsequent 3 elements of the vector Yi are skipped and it is determined whether the fourth element of the vector still has value T and is preceded by only T: if the response is positive, then the pair of coordinates of the extracted route relative to the index 4 of the vector Yi. is stored in the final polyline.

If, on the other hand, the response is negative, then it is determined to which vector Yi the element which interrupts the series of T of the vector Yi belongs, that is to say, the first occurrence of Ό’ in that vector and the process is repeated changing the vector Yi with the vector Yi selected, changing also Dei and K, until the end of all the pairs of coordinates of the route extracted.

Returning to the example, having the first 30 elements of the vector Yi all equal to 1 , 10 pairs of coordinates of the extracted route will be inserted in the final polyline, 1 in each 3 equal to the value of Ki, changing the vector Yi with another since the 31 st element of the vector will be equal to O’, so the relative pair of coordinates no longer belongs to the level of difficulty Dei. These results can be both displayed on the display unit 4 as static information, that is to say, the user views the difficulty of the curve and takes this into account during driving, or they can be stored or, also, integrated inside existing polylines, or also shared or made available to navigation applications to improve the reliability.

Since the density of the polyline is variable, there is no risk of significantly increasing the computational load of any other navigation systems.

The effect which is obtained is an optimised adherence of the quantity of data to the needs of the user, with a more precise information system in the presence of dangerous curves and more bland during straight roads or gentle curves.

As is apparent from the above description, the advantage of the system 1 described is that of providing support to the driver, providing information in advance on the difficulty of the curve close to the driver’s position and, therefore, improving the modes of travel.

The system 1 generates information on the road route in an offline mode, so is to be considered preparatory for a further application which uses these services to support the functions described in the aims.

This invention is described by way of example only, without limiting the scope of application, according to its preferred embodiments, but it shall be understood that the invention may be modified and/or adapted by experts in the field without thereby departing from the scope of the inventive concept, as defined in the claims herein.

Claims

1 . System (1 ) for calculating curvature values (Vd) of a curve (Ci) in a road route comprising:

storage means (21 ) for storing at least one satellite cartographic map containing said road route;

a logic control unit (2), operatively connected to said storage means (21 ) for retrieving said at least one satellite cartographic map, said system (1 ) being characterized in that said logic control unit (2) is equipped with processing means (22, 23) of said at least one satellite cartographic map, comprising a calculation program which implements an algorithm which provides as output

a calculation of the curvature values (Vd) at each point of said road route, and

the association to said curvature values (Vd) of a data structure whose distribution is variable corresponding to said calculated curvature values (Vd).

2. System (1 ) according to the preceding claim, characterized in that said data structure is a polyline comprising a plurality of points having a variable density so that, in correspondence with curvature values (Vd) that exceed a predetermined threshold (Xd), the points of said polyline have a density higher than the density of the points of the polyline corresponding to curvature values (Vd) lower than said predetermined threshold (Xd).

3. System (1 ) according to any one of the preceding claims, which can be installed on a vehicle, characterized

in that it comprises a driving unit (5) connected to said logic control unit (2), and connectable to the operating unit of said vehicle,

in that said driving unit (5) receives as input the physical parameters, the forward speed value and the geographical positioning coordinates of said vehicle, and

in that said driving unit (5) is configured for

processing said calculated curvature values (Vd), said physical parameters of the vehicle, forward speed value of said vehicle and geographical positioning coordinates of the vehicle, and

for sending corresponding control signals to said control unit, so as to brake or accelerate the forward speed of said vehicle.

4. System (1 ) according to claim 2 or 3, characterized in that it comprises a display unit (4) for displaying said road route on said satellite cartographic map, in the form of polylines associated with it.

5. Operation method of a system (1 ) for calculating curvature values (Vd) of a curve (Ci) in a road route according to any one of the preceding claims, characterized in that it comprises the following steps: a. generating an image (S) associated to a predetermined portion of a satellite cartographic map, starting from a retrieved satellite cartographic map, containing a road route;

b. if said image (S) has not a predetermined resolution, generating a NxN matrix of sub-images (SN,N);

c. if the step b has been carried out, blending said sub-images (SN,N) of said NxN matrix, to obtain a further image SF, different from said image S;

d. removing one or more indicators, such as labels or points of interest present on said retrieved satellite cartographic map, from said image (S), or from said further image (SF);

e. converting said image (S), or said further image (SF) into a binary image (SB);

f. generating a skeletonized image (SK) starting from said binary image (SB);

g. detecting and removing one or more further road routes with respect to said road route of interest, present in said skeletonized image (SK);

h. generating spatial coordinates for each point of said road route;

/. analysing the gradients in each of said points;

j. classifying the curves present in said road route;

k. generating a polyline having a variable density.

6. Method according to the preceding claim, characterized in that said binary image (SB) generated in said step f. comprises a white pattern comprising a plurality of pixels in which each pixel uniquely identifies a point of the center of the road of said road route, and a pair of geodesic coordinates of the point itself is associated to each pixel.

7. Method according to the preceding claim, characterized in that in said steps h. and /. said processing unit (23) calculates the successive derivatives on the set of said geodesic coordinates extracted in said step f. to obtain the curvature value (Vci) in each point of said road route.

8. Method according to the preceding claim, characterized in that in said step j. thresholds (X_n) of said curvature values (Vd) calculated in said steps h. and /. are established and difficulty levels (Dd) are associated with said thresholds (Xn).

9. Method according to the preceding claim, characterized in that in said step k., said processing unit (23) receives as input

the set of said geodesic coordinates belonging to said road route, the classification data of said step j.

and it performs a variable sampling of the curves (Ci) present in said road route, in particular it makes a selection of a sub-set of coordinates of points belonging to a determined curve (Ci).