A method for processing seismic images to obtain a rgt image

Abstract

It is disclosed a method and computing device for processing a seismic image obtained from seismic measurements, said seismic image extending along at least one horizontal direction and one vertical direction, and comprises a plurality of pixels where each pixel is associated with a seismic value, said method comprising steps of: dividing the seismic image along the horizontal direction in order to obtain a plurality of columns, initializing, for each column, a plurality of segments, where each segment is determined from the seismic values of pixels contained in the column, generating a plurality of patches by iterative merging of segments initialized for adjacent columns, the iterative merging being performed under the constraint that a determined patch cannot comprise portions extending above and below a second patch, along the vertical direction, and generating a relative geological time, RGT, image based on the seismic image and the generated patches.

Description

[0001] A METHOD FOR PROCESSING SEISMIC IMAGES TO OBTAIN A RGT IMAGE

[0002] TECHNICAL FIELD

[0003] The present disclosure relates to the processing of seismic images of a geological formation in order to obtain a chrono-stratigraphic representation of the geological formation, a.k.a. relative geological time, RGT, image of the geological formation.

[0004] BACKGROUND ART

[0005] It is known, especially in hydrocarbon exploration, to determine the position of reservoirs from the results of geophysical measurements carried out from the surface or in well bores.

[0006] According to the technology of reflection seismology, these seismic measurements involve emitting a wave (e.g. acoustic wave) into the subsurface and measuring a signal comprising a plurality of echoes of the wave on geological structures being investigated. These structures are typically surfaces separating distinct materials, faults, etc. Other measurements may be carried out at well bores.

[0007] Chrono-stratigraphic analysis is very important to understand basin evolution, predict the sedimentary facies distribution for both hydrocarbon exploration and development, or other applications such as carbon storage. This analysis is based on the fundamental assumption that seismic reflectors are surfaces of chrono- stratigraphic significance. This assumption implies that an individual seismic reflector is a “time-line” through a depositional basin that represents a surface of the same geological age (i.e. an isochronous surface in geological time).

[0008] A seismic image (or seismic section) comprises a juxtaposition in a volume of sampled one-dimensional signals referred to as seismic traces. In the seismic image, the value of a pixel (i.e. a voxel for 3D images) is proportional to the seismic amplitude represented by seismic traces.

[0009] Computing a chrono-stratigraphic representation of a seismic image often requires determining seismic horizon surfaces of the seismic image, wherein a seismic horizon surface corresponds to an estimated isochronous surface of the geological formation. Such seismic horizon surfaces can be used to determine an RGT image of the geological formation, i.e. an image in which each pixel provides an estimated geological age for the portion of the geological formation represented by said pixel. The RGT image is referred to as “relative” because the purpose of the RGT image is mainly to be able to compare the estimated geological ages of different pixels, in order to e.g. identify portions of the geological formation that have the same estimated geological age. Also, in practice, it is usually not possible to estimate an absolute geological age of any given portion of the geological formation.

[0010] It is known from EP3978960A1 a method for processing a seismic image to construct a plurality of seismic horizons, said method comprising determining a seismic dip image comprising local seismic dips representative of the local gradient of the seismic values of the seismic image, initializing a number of horizons from respective seeds, where a seed is a point of a seismic trace that satisfy a predetermined criterion, and iteratively modifying the horizon surfaces to progressively increase alignment between local orientations of the seismic horizon surfaces and the corresponding local seismic dips.

[0011] It is also known from the publication of Xinming Wu “Horizon volumes with interpreted constraints”, in Geophysics, February 2015, 80(2):IM21-IM33 DOI:10.1190 / geo2014- 0212.1 , a method for constructing seismic horizons aligned with reflectors in a 3D seismic image, where horizons are generated at once by solving partial differential equations derived from seismic normal vectors.

[0012] The horizons generated thanks to these methods respect the seismic reflectors present within the seismic image. However, the generated horizons may not be temporally coherent, i.e. some horizons may cross each other, and thus comprise portions extending both over and below another horizon. This cannot happen in reality since the horizons are formed chronologically, in sequence, and hence must not be present in a RGT image derived from a seismic image. Hence, the generated horizons may induce errors in the RGT images that are derived therefrom, and then in the geological models which are built for reproducing and simulating a studied geological reservoir.

[0013] SUMMARY

[0014] The present disclosure aims at improving the situation. In particular, the present disclosure aims at overcoming at least some of the limitations of the prior art discussed above, by proposing a solution for extracting patches from a seismic image that can be used as constraints in a process of generating horizons in a RGT image and which enables that the generated horizons do not cross each other.

[0015] According to a first aspect, a computer-implemented method for processing a seismic image obtained from seismic measurements performed on a geological formation is disclosed, said seismic image extending along at least one horizontal direction and one vertical direction, and comprises a plurality of pixels where each pixel is associated with a seismic value, said method comprising steps of: dividing the seismic image along the horizontal direction in order to obtain a plurality of columns, initializing, for each column, a plurality of segments, where each segment is determined from the seismic values of pixels contained in the column, generating a plurality of patches by iterative merging of segments initialized for adjacent columns, the iterative merging being performed under the constraint that a determined patch cannot comprise portions extending above and below a second patch, along the vertical direction, and generating a relative geological time, RGT, image based on the seismic image and the generated patches.

[0016] In embodiments, the generation of patches comprises determining adjacent patches sharing a common edge, assigning a weight to each common edge, and wherein the iterative merging of the segments is performed according to a priority order determined from the weights of the common edges.

[0017] In embodiments, the weight associated to a common edge is computed from at least one of the following: an amplitude of at least one segment among the two segments, an amplitude gap between the two segments, a size of the common edge between the two segments.

[0018] In embodiments, the iterative merging of segments is performed until a stop criterion is fulfilled, the stop criterion comprising at least one among the following criteria: a determined number of pairs of segments have been merged, an amplitude value of at least one segment among two segments considered for merging is below a predefined amplitude threshold, an amplitude gap value of amplitudes between the two segments considered for merging exceeds a predefined gap threshold, a size value of at least one segment among the two segments considered for merging is below a predefined size threshold.

[0019] In embodiments, the stop criterion comprises a weight of a common edge between two segments considered for merging being below a determined threshold.

[0020] In embodiments, the method further comprises merging unmerged segments based on relaxed stop criteria.

[0021] In embodiments, two segments are merged if all merged segments form a graph without edge intersection, the segments being nodes of the graph and edges linking two nodes representing a merge between two segments.

[0022] In embodiments, two segments are merged if they do not belong to the same column, or a patch is merged with a segment if the match does not comprise any segment belonging to the same column as the segment considered for merging.

[0023] In embodiments, the merging of two segments is forbidden if a vertical distance between the segments exceeds a predetermined threshold.

[0024] In embodiments, the initialization of the plurality of segments for each column comprises: determining a plurality of seed segments based on intersections between seismic reflectors identified from the seismic values of the pixels of the seismic image and a seismic trace of the column extending parallel to the vertical direction, and propagating laterally each seed segment to obtain a corresponding segment. In embodiments, the seismic trace is a central seismic trace of the column.

[0025] In embodiments, generating the RGT image comprises is performed under the constraint that each patch corresponds to an isochronous geological time surface.

[0026] According to another object, it is disclosed a computer-program product having stored thereon instructions which, when executed by a processor, causes the processor to perform the method according to the description above. According to another object, it is disclosed a computing device, comprising at least a processor and a memory, the computer being configured for implementing the method according to the description above.

[0027] The method enables generating a plurality of patches which can then be used as constraints for an algorithm of generation of RGT image from a seismic image. The resulting RGT image is rendered more reliable and compliant with geology, which enable then better understanding and more accurate simulations (for instance of fluid flow) of reservoir models that are derived from the RGT image.

[0028] BRIEF DESCRIPTION OF DRAWINGS

[0029] The present disclosure will be better understood upon reading the following description, given as an example that is in no way limiting, and made in reference to the figures which show:

[0030] Figure 1 represents a seismic image divided into a plurality of columns

[0031] Figure 2a represents examples of segments defined in a seismic image and the potential connections between the segments,

[0032] Figure 2b represents the patches generated from the segments of figure 2a.

[0033] Figure 3 schematically represents an example of common edge between two segments.

[0034] Figure 4 represents an example of graph of merged segments.

[0035] Figure 5 represents an example of forbidden merging between two patches each comprising a plurality of segments,

[0036] Figure 6 is a flow chart showing the main steps of a method according to an embodiment,

[0037] Figure 7 schematically represents a computing device according to an embodiment.

[0038] DESCRIPTION OF EMBODIMENTS

[0039] As discussed above, the present disclosure relates inter alia to a method 100 for processing seismic images.

[0040] A seismic image represents a picture of the subsoil arising from a seismic exploration survey. The seismic image comprises at least two dimensions which may comprise at least one horizontal dimension (which usually uses a distance scale, expressed e.g. in meters), i.e. one dimension along a horizontal direction (y on figure 1 ), and one vertical dimension (which usually uses a distance scale or a time scale, expressed e.g. in seconds), i.e. one dimension along a vertical direction (x on figure 1 ). Hence, the seismic image may correspond to a 3D seismic image (extending along two horizontal directions and one vertical direction) or to a 2D seismic image (extending along one horizontal direction and one vertical direction).

[0041] The seismic image is composed of pixels which may be 2D in case of a 2D seismic image or 3D (voxels) in case of a 3D seismic image. The pixels are regularly distributed according to a horizontal pitch on each horizontal direction and a vertical pitch on the vertical direction.

[0042] The seismic image comprises, along each horizontal direction, a number of columns of pixels which is referred to as “seismic traces”. The number of seismic traces is equal to the quotient of the dimension along this horizontal direction, divided by the horizontal pitch along this direction. The seismic image further comprises a number of pixels per seismic trace which is equal to the quotient of the vertical extension divided by the vertical pitch.

[0043] Each pixel is associated with a seismic value which may be a gray value, for instance between 0 and 255. More generally, the seismic values may be coded in 8, 16, 32 or 64 bits, with or without a sign. Thus, for 8 bits without sign, the seismic value associated with a pixel can range between 0 and 255, whereas when the values have a sign they can range between -127 to +127. Each seismic value is representative of the amplitude of the seismic signal measured for the position associated to the corresponding pixel. Thus, in what follows, the seismic value of a pixel may also be called amplitude.

[0044] Figure 1 represents an example of seismic image. The seismic image represented by figure 1 is 2D and may correspond to a 2D seismic image or to a 2D section of a 3D seismic image, in a vertical plane comprising one of its horizontal dimensions. As can be seen in figure 1 , the seismic values highlight the composition of the geological formation, since high amplitude seismic values are usually associated to strong seismic reflectors, which are usually located at the interfaces between geological layers having different acoustic impedances. Figure 6 represent schematically the main steps of an exemplary embodiment of a method 100 for processing a seismic image.

[0045] The processing method may be carried out by a computer system 1 , schematically represented in figure 7. In preferred embodiments, the computer system 1 comprises one or more processors 10 (which may belong to a same computer or to different computers) and storage means 20 (magnetic hard disk, optical disk, electronic memory, or any computer readable storage medium) in which a computer program product is stored, in the form of a set of program-code instructions to be executed in order to implement all or part of the steps of the processing method 100. Alternatively, or in combination thereof, the computer system can comprise one or more programmable logic circuits (FPGA, PLD, etc.), and / or one or more specialized integrated circuits (ASIC), etc., adapted for implementing all or part of said steps of the processing method 30. In other words, the computer system comprises a set of means configured by software (specific computer program product) and / or by hardware (processor, FPGA, PLD, ASIC, etc.) to implement the steps of the processing method 30.

[0046] As illustrated in figure 6, the processing method comprises a step 110 of dividing the seismic image along the horizontal direction (or directions when the seismic image is in 3D) in order to obtain a plurality of columns C, where each column has a crosssection comprising a plurality of pixels. The columns obtained at step 110 are distinct from the seismic traces mentioned above. In particular, the columns obtained at step 110 include a plurality of seismic traces. Moreover, the dimension along the horizontal direction(s) of each column may be pre-determined, or it may be adjusted (for instance by a user) according to the seismic image to be processed. In particular, when the seismic image exhibits seismic reflectors which are flat, the dimension along the horizontal direction(s) of the columns may be greater than when the seismic reflectors are strongly folded, since in the latter case there is increased requirement for precision.

[0047] Moreover, the columns are defined such that each column partially overlaps the adjacent column(s). Two 2D columns may overlap by a 2D band of pixels extending over the vertical dimension and comprising one or more pixels in the horizontal direction. In the case shown in figure 3, two 3D columns may overlap by a parallelepiped extending over the vertical dimension, having a dimension along a horizontal direction that is equal to the dimension of the overlapping columns (along the z axis in the figure) and having a dimension of one or more pixels along the other horizontal direction (y axis in the figure). The zone of overlap between two adjacent columns is called recovery area.

[0048] The method then comprises a step 120 of initializing, for each column, a plurality of segments S. The segments may also be referred to as micro-patches. Each segment corresponds to a set of pixels defining a laterally bounded surface. Each segment extends within the boundaries of the column in which it has been defined.

[0049] The segments are defined from the seismic values of pixels contained in the column. More specifically, for a given column, a segment seed is initiated 121 for a plurality of seismic events, or seismic reflectors, that are apparent in the column. In embodiments, the segments seeds may be defined from the seismic values of pixels of one seismic trace comprised in the column, for instance the central seismic trace of each column. One segment seed may be initialized each time the considered seismic trace crosses a seismic reflector of the seismic image. This may be performed by processing the seismic values of the pixels of the seismic trace to detect sudden intensity change in the values of successive pixels along the vertical direction.

[0050] Accordingly, at the time of initialization, the horizontal dimension of each segment seed is that of the seismic trace from which it has been generated. In embodiments, step 120 further comprises propagating 122 each segment seed within the column so that it reaches the lateral (horizontal) boundaries of the column and forms a segment or micro-patch. The propagation of a segment seed may be performed by searching for pixels of adjacent seismic traces that satisfy a predetermined criterion of similarity with the seed. For instance, the lateral search may be performed by correlating an adjacent seismic trace with vertically translated copies of the seismic trace comprising the initialized segment seed in order to identify the vertical translation that optimizes the similarity (e.g. maximum correlation value and / or correlation value above a predetermined threshold) with the seismic trace comprising the seed. The pixels around a segment seed that satisfy the similarity criterion with said seed, and their respective positions in the column, define the segment. On figure 2a are represented a plurality of segments which have been defined for three adjacent columns, where each segment is associated to a respective number.

[0051] The processing method then comprises a step 130 of generating, from the segments S, a plurality of patches P, where each patch is formed by a plurality of adjacent segments, i.e. a plurality of segments issued respectively from adjacent columns C. In the present disclosure “adjacent columns” implies that the columns are contiguous. In the example of figure 2a, the left column is adjacent to the middle column but not with the right column.

[0052] Step 130 includes iterative merging of segments obtained at step 120 and which share a common edge, under a constraint that the generated patches cannot cross each other, i.e. a determined patch cannot comprise portions extending both above and below another patch, considered in the vertical direction.

[0053] Step 130 may comprise a preliminary step 131 of selecting segments candidate for merging, where the selected segments fulfill at least one determined condition. For instance, step 131 may comprise selecting segments having a minimum size, i.e. for instance having a size, measured in the horizontal direction(s), which is above a determined threshold. The determined threshold may be expressed relative to the size of the columns along the same horizontal direction(s). According to a non-limiting example, step 131 may comprise selecting segments having a size superior or equal to at least 25% of the size of the columns, in the horizontal direction(s).

[0054] Another condition may rely on the discontinuities present within the segment. Discontinuities may occur within a segment due to, for instance, poor quality of the seismic image, corresponding for instance to a high level of noise. Discontinuities within a segment may also occur when a fault extends within the column in which the segment has been defined. The selected segments may exhibit a number or amount of discontinuities that is below a determined threshold.

[0055] Whether this preliminary selection step 131 is performed or not, step 132 comprises determining all pairs of adjacent segments among the set of segments that have been generated (and, as the case may be, selected), i.e. all pairs sharing a common edge. With reference to figure 3, a common edge E is defined between two adjacent segments denoted S1 and S2 in the figure, when the two segments share a number of common pixels, in the recovery area R between the two columns in which the segments are formed, that exceeds a determined threshold.

[0056] In embodiments, step 132 further comprises building a graph of neighboring segments, formed of a plurality of nodes and connections between the nodes, where the nodes of the graph are formed by segments, and the connections between the nodes are formed by common edge defined as explained above. With reference to figure 2a is shown an example of a graph of neighboring nodes obtained from the segments displayed in the seismic image on the left-hand part of the figure. As one can notice from the exemplary graph, one segment may share a common edge with a plurality of segments. This is the case of segment 35 in figure 2a which shares a common edge with both segments 25 and 26 of the same column.

[0057] Step 130 of generating patches then comprises a step 133 of iteratively merging neighboring segments, i.e. segments having a common edge, in order to form a plurality of patches. Step 133 of merging neighboring segments may use the graph pf neighboring segments obtained at step 132 as input. As can be readily understood, as step 133 comprises iterative steps of merging neighboring segments, each iteration may include either merging two segments (especially at the beginning of the iterations) or merging one patch, already formed by a plurality of merged segments, with another segment, which shares a common edge with one of the merged segments of the patch.

[0058] In embodiments, the merging of segments having a common edge may be performed by priority order according to a weight associated with each common edge. According to non-limiting example, the weight associated with each common edge may be indicative of a level of continuity, in amplitude (i.e. seismic value) and / or geometry, of the merged segments. Furthermore, more weight may be given to segments of high amplitude, i.e. of high seismic value, in order to merge in priority the segments corresponding to the strongest reflectors.

[0059] The merging is thus performed by decreasing value of the weight associated with segments, where the weight may be computed from at least one of the following quantities, or any combination thereof: an amplitude of at least one segment among the two segments, and preferably an amplitude of the two segments or a quantity derived therefrom, an amplitude gap between the two segments, a size of the common edge between the two segments, i.e. a length of the pixels in common at the recovery area between the two segments, if the seismic image is in 2D, or a surface of the pixels in common, if the seismic image is in 3D.

[0060] The amplitude of a segment may be computed as the average amplitude, i.e. seismic value, of the pixels belonging to the segment. The amplitude gap between the two segments may be computed as a difference between the average amplitudes of the two segments.

[0061] According to a non-limiting example, the weight of a common edge of two adjacent segments may be computed as follows: w = — a1c1+ a2c2+ z3c3)

[0062] Where cLis a criterion and aLa ponderation weight, which may be defined as follows: c1 is the maximum average amplitude value between the two segments p1 and p2 sharing a common edge and considered for merging ; c2 is the size of the common edge between p1 and p2, c3 is a criterion evaluating an amplitude gap between the two segments, which may be defined as follows:

[0063] Where amp; is the average amplitude of segment pi.

[0064] The ponderation weights may be positive values.

[0065] In embodiments, the iterative step 133 of merging pairs of segments further comprises building a graph of merged segments, comprising a plurality of nodes and edges between the nodes, where the segments are nodes of the graph and edges linking two nodes represent a merge between two segments. Furthermore, the graph is defined such that the disposition of the nodes respects the disposition of the segments in the seismic image. In particular, segments belonging to a same column are represented by vertically aligned nodes, and the order of the segments along the vertical direction in the seismic image is maintained within the alignment of nodes along the vertical direction. Furthermore, the relative position of the columns according to a horizontal direction is preserved, i.e. all aligments of nodes corresponding to respective columns are positioned in the same order as the columns from which they originate.

[0066] At each iteration of step 133, the merging between two segments is allowed if the updated version of the graph, comprising the newly merged segment, comprises no intersection, i.e. no intersecting edge. According to an example represented in figure 4, two alignments of nodes A1...A5 and B1...B5 are represented where each alignment correspond to a column. Let us assume that a first pair of segments A2 and B3 has been merged. Then the next common edge to be considered for merging, according to the priority order, is a common edge between A1 and B4. Such merging will be forbidden for an edge linking A1 and B4 would cross the pre-existing edge between A2 and B3. Alternatively, is the common edge considered for merging was between A3 and B4, then the merging of these nodes would be allowed.

[0067] According to this constraint, it is ensured that a patch, which is formed by the merging of a plurality of segments, does not cross another patch, i.e. it cannot comprise portions extending both above and below another patch. As exposed in more details below, as the generated patches serve as constraints for building horizons in a RGT image, this ensure that the built horizons are not going to cross one another and hence that the chronological sequence of formation of the horizons is preserved.

[0068] According to another constraint, and as represented in figure 5, the merging of two segments is allowed if the segments do not belong to the same column. In figure 5, the segments of the two patches corresponding to the same column are shown by arrows. This constraint is of particular interest after a plurality of iterations of the merging step, when patches comprising a plurality of merged segments have been formed and a current merging step is about merging two patches, each comprising a plurality of segments. In that case, two patches cannot be merged if they include a segment formed from a common column.

[0069] In embodiments, according to another constraint, the merging of two segments is allowed only if the merging does not create a vertical discontinuity. In other words, if the two segments are vertically offset from one another by a distance exceeding a predetermined threshold, the merging is forbidden. In the case of segments that result from previous mergings and that are composed of a plurality of initial segments, corresponding to a plurality of respective columns as represented in figure 5, said condition may be verified by verifying only the relative vertical positions of the parts of the segments considered from merging that belong to neighboring columns, and which thus form part of the edges of the segments. If at least one vertical gap exceeding the predetermined threshold is determined, the merging between the segments is forbidden.

[0070] The iterations are performed until reaching a stop criterion. The stop criterion may be a minimum value on the weight of the common edges that are considered for merging. Alternatively, the stop criterion may be a limit value on any of the following: an amplitude value of at least one segment among two segments considered for merging, and possibly of the two segments, is below a determined threshold, an amplitude gap value of amplitudes between the two segments considered for merging exceeds a determined threshold, a size value of at least one segment among the two segments considered for merging is below a determined threshold.

[0071] In embodiments, a plurality or all among the stop criteria defined above may be applied, and the iterations are performed until one stop criterion is fulfilled.

[0072] In embodiments, the method may then comprise reiterating step 133 of segments merging by relaxing 134 the constraints considered for the merging, i.e. changing the thresholds considered for the stop criteria considered above. For instance, a user may examine the patches that have been generated and may allow to continue the step 133 of segments merging by relaxing the constraints, for instance by setting new thresholds value for the stop criteria. Also, when two patches have not been merged because of segments located in a common column, the user may manually remove a segment from one of the patches to allow the fusion of the patches, if he considers (e.g. from the study of the seismic image) that the two patches correspond to the same geological event.

[0073] With reference to figure 2b are shown the patches generated from the segments defined in figure 2a. One can see that a plurality of patches overlapping the three columns shown in figures 2a and 2b have been defined. Furthermore, the graph of merged segments displayed on the right-hand side of figure 2b shown that the merging between segments 25 and 35 has been forbidden, even though these segments shared a common edge, because it would have generated two crossing edges (between 25 and 35 and between 26 and 35).

[0074] The method then comprises a step 140 of generating a RGT image from the seismic image, using the generated patches as constraints during said generation. A RGT image is an image derived from the seismic image, where each pixel is associated to a respective relative geological time T. The patches obtained as step 130 are used as constraints on the RGT value such that each patch corresponds to a constant RGT value. The skilled person knows methods for generating a RGT image from a seismic image. According to a non-limiting embodiment, step 140 may comprise implementing the iterative algorithm disclosed the publication discussed above from Wu “Horizon volumes with interpreted constraints”, where the RGT image computed at each iteration n is updated by replacing the RGT values of the voxels corresponding to a common patch computed according to the method above by a constant RGT value. Said constant RGT value may be computed as the mean RGT value among the RGT values computed at the previous iteration n-1 over the pixels belonging to the considered patch.

[0075] The obtained RGT image may be used to start the analysis of the geological formation.

[0076] It is emphasized that the present invention is not limited to the above exemplary embodiments. Variants of the above exemplary embodiments are also within the scope of the present invention.

Claims

CLAIMS1. A computer-implemented method for processing a seismic image obtained from seismic measurements performed on a geological formation, said seismic image extending along at least one horizontal direction and one vertical direction, and comprises a plurality of pixels where each pixel is associated with a seismic value, said method comprising steps of: dividing (110) the seismic image along the horizontal direction in order to obtain a plurality of columns (C), initializing (120), for each column (C), a plurality of segments (S), where each segment is determined from the seismic values of pixels contained in the column, generating (130) a plurality of patches (P) by iterative merging of segments (S) initialized for adjacent columns, the iterative merging being performed under the constraint that a determined patch cannot comprise portions extending above and below a second patch, along the vertical direction, and generating (140) a relative geological time, RGT, image based on the seismic image and the generated patches.

2. The method according to claim 1 , wherein the generation of patches (130) comprises determining (131 ) adjacent patches sharing a common edge (E), assigning a weight to each common edge, and wherein the iterative merging (133) of the segments is performed according to a priority order determined from the weights of the common edges.

3. The method according to any one of the preceding claims, wherein the weight associated to a common edge (E) is computed from at least one of the following: an amplitude of at least one segment among the two segments, an amplitude gap between the two segments, a size of the common edge between the two segments.

4. The method according to any one of the preceding claims, wherein the iterative merging of segments is performed until a stop criterion is fulfilled, the stop criterion comprising at least one among the following criteria: a determined number of pairs of segments have been merged, an amplitude value of at least one segment among two segments considered for merging is below a predefined amplitude threshold, an amplitude gap value of amplitudes between the two segments considered for merging exceeds a predefined gap threshold, a size value of at least one segment among the two segments considered for merging is below a predefined size threshold.

5. The method according to claim 4 combined with claim 2, wherein the stop criterion comprises a weight of a common edge between two segments considered for merging being below a determined threshold.

6. The method according to claims 4 or 5, further comprising merging unmerged segments based on relaxed stop criteria (140).

7. The method according to any one of the preceding claims, wherein two segments are merged if all merged segments form a graph without edge intersection, the segments being nodes of the graph and edges linking two nodes representing a merge between two segments.

8. The method according to any of the preceding claims, wherein two segments are merged if they do not belong to the same column, or a patch is merged with a segment if the match does not comprise any segment belonging to the same column as the segment considered for merging.

9. The method according to any of the preceding claims, wherein the merging of two segments is forbidden if a vertical distance between the segments exceeds a predetermined threshold.

10. Method according to any one of the preceding claims, wherein the initialization of the plurality of segments for each column comprises:determining a plurality of seed segments based on intersections between seismic reflectors identified from the seismic values of the pixels of the seismic image and a seismic trace of the column extending parallel to the vertical direction, and propagating laterally each seed segment to obtain a corresponding segment.11 . The method according to claim 10, wherein the seismic trace is a central seismic trace of the column.

12. The method according to any one of the preceding claims, wherein generating (140) the RGT image comprises is performed under the constraint that each patch corresponds to an isochronous geological time surface.

13. A computer-program product having stored thereon instructions which, when executed by a processor (10), causes the processor to perform the method according to any of the preceding claims.

14. A computing device (1 ), comprising at least a processor (10) and a memory (20), the computer being configured for implementing the method according to any of steps 1 to 12.

Images