Computer script for processing images and use thereof in petrographic image determination method

Abstract

The present invention relates to a computing script designed to extract contour contrasts from images to facilitate methods for lithofacies image determination. The present invention presents a tool for highlighting the heterogeneity of rocks to identify textural patterns and typical structures related to sedimentary facies. The "Canny edge detection" algorithm and settings in the Python computer language environment are used to extract the contour contrasts observed in the cross-sections of the captured rock images. Other results achieved by the present invention include the removal of artifacts, the superposition of images, and the quantification of edge contrasts. These results have been incorporated into the methods for lithofacies image determination. In addition, the use of the resulting products in well-logging facies prediction models, deliverability, and interwell correlations has a wide range of prospects. Thus, the script developed provides important information for the oil industry.

Description

Technical Field

[0001] This invention relates to computational scripts for image processing that enable the highlighting of rock texture and structural variations by extracting edge contour contrast, removing artifacts (non-geological features present in image profiles), and overlaying images, as well as enabling the quantification of edge contrast. These products can be used in methods for determining lithofacies from image profiles. Background Technology

[0002] Electronic and ultrasonic imaging profiling tools capture a wealth of high-resolution data around the wellbore. One purpose of these profiles is to identify and interpret geological events by correlating measurements performed subsurface with geological data. This information enables detailed descriptions of stratigraphic geological characteristics, facilitating sedimentological analysis, stratigraphic analysis, structural analysis, geomechanical analysis, petrophysical analysis, and reservoir characterization.

[0003] Image profiles are highly susceptible to artifacts, some of which can be minimized or corrected if identified promptly. Therefore, control measures during image profile acquisition are crucial for ensuring the best possible data quality and increasing the reliability of stratigraphic interpretation and assessment.

[0004] High-quality profile images combined with rock data provide an excellent tool for lithofacies and structural analysis. Integrating this data with stratigraphic test results into geological models adds significant value.

[0005] Image-based petrographic studies consider the texture and sedimentary structure of rocks observed in images and generate visual associations between lithofacies and image lithofacies based on geological knowledge of the region and interpreter experience. The petrographic descriptions used (lithofacies and image lithofacies) generally follow those proposed by Terra et al. de Rochas Bacias Sedimentares Brasileiras.Boletim de The classification by da Petrobras (BGP) in mid-2010. For a more detailed explanation, as proposed in Reading, Sedimentary environments: process, facies and stratigraphy, 3rd edition, Blackwell, Oxford, 1996, additional analysis of the diagenesis, volume and type of clay minerals, and primary structure will be necessary.

[0006] Rock × profile integration helps characterize the reservoir potential around the wellbore. Formation tests provide information on reservoir flow models at various distances outside the wellbore. The integration of these data increases the accuracy of geological and reservoir flow models, thereby enhancing oilfield development and production management.

[0007] Due to the high costs associated with extracting lithofacies evidence, the use of image profiles is crucial in lithofacies interpretation. For this purpose, it is equally necessary that the images have good quality and resolution, optimized through geological tracking during analysis.

[0008] To find ways to improve image profiles used in petrographic interpretation, thereby adding value to the final product and giving geological models reliability.

[0009] Image processing that attempts to extract contrasting contours from observed edges corresponds to important tools for generating products that contribute to the characterization of lithofacies in images and the correlation between rock × section maps. These edge contrasts represent variations in texture and structure within lithofacies, as well as megapores or artifacts ranging from megapores to gigapores.

[0010] As shown in Choquete, PW, and Pray L's 1970 work, *Geologic Nomenclature and Classification of Porosity in Sedimentary Carbonates*, *The American Association of Petroleum Geologists Bulletin*, Vol. 5, pp. 207–250, megapores are understood as pore sizes between 0.4 cm and 25.6 cm, while gigapore sizes are presented as in Menezes de Jesus, C, Compan, AL, and Surmas, R, 2016, in *Petrophysics*, Vol. 57, pp. 620–637, *Permeability Estimation Using Ultrasonic Borehole Image Logs in Dual-Porosity Carbonate Reservoirs*, where gigapore sizes are greater than 25.6 cm.

[0011] It is important to emphasize that rocks are a direct and irreplaceable source of information about stratigraphy. However, it is crucial to obtain methods for extracting as much information about stratigraphy as possible. This information can be used for rock × profile calibration, different types of correlations, and as a quality control / standard (beacon) in work involving prediction.

[0012] Given the need to reduce operating costs and extract as much information as possible from image profiles, Bal et al., in their 2001 work, attempted to apply a method for determining lithofacies based on image profiles. Since then, the method has been improved and applied to wells analyzed by Petrobras.

[0013] In their 2018 work, Fioriti and Mello Jr. developed a computational script programmed in Python that is capable of extracting edge contrast contours observed in any type of image, highlighting the inhomogeneity of rocks, for the purpose of identifying typical textures and structural patterns that may be associated with sedimentary rock facies.

[0014] Given the importance of identifying texture and structural patterns in lithofacies interpretation, and the fact that some sedimentary structures are more apparent in image profiles, fault scan images, or evidence, it is important to recognize the need to improve the way these texture variations and structural elements are visualized in any type of image.

[0015] Conventional image editing programs have tools to highlight edge contours in images. However, these programs were not developed with the intention of generating value for the oil industry—the purpose of this invention. Therefore, they do not provide the same results and products as the scripts developed in this invention. A key innovation of this invention is the quantification of edge contrast.

[0016] Commercial software such as Techlog, IP, and Geolog, used for processing and interpreting image profiles, do not have image processing modules that provide the results provided by the scripts developed in this invention.

[0017] The “Canny edge detection” algorithm (Canny, JF; 1986) exists in the OpenCV library, which is used in programming software written in Python. To implement this invention, the OpenCV library was used, and an attempt was made to extract the contrast of edge contours observed in images (i.e., acoustic and electrical resistivity image profiles, tomographic or evidence photographs, lateral samples, and thin sections, as presented in the 2018 work of Fioriti and Mello Jr.) with the help of personalized algorithm parameterization.

[0018] The improvements of this invention enable other objectives, such as: removing tool marks (artifacts) that make the texture and structure of rocks difficult to visualize; overlaying the original image with detected edges, which further highlights the heterogeneity of the rock; and quantifying the edge contrast identified by depth. This information has various potential applications, ranging from correlation with lithological variations to correlation with data from other profiles and stratigraphic tests.

[0019] Document WO2009126881A2 discloses a method for generating three-dimensional (3D) models (known as numerical pseudo-kernels) of rocks and boreholes. This method uses a complete circular image of the wellbore, a digital rock image, and an algorithm with continuous variables formed within a multi-point statistical (MPS) range to reproduce a 3D pseudo-kernel for recording intervals in which the true kernel is not removed, but rather presents borehole images obtained from the recording. Therefore, the applied method does not use the "Canny edge detection" algorithm in a Python computing environment, such as the Python computing environment of this invention.

[0020] Document US20190338637A1 discloses a method for determining the properties of geological strata based on optical images of rock samples taken from strata. The image comprises multiple pixels, and the method includes defining windows in the image, each window comprising a predetermined number of pixels and each window having a predetermined shape. The method also includes, for each window, extracting a representative rock impression value for that window. The rock impression includes indicators used to characterize the texture of the window. The method further includes classifying the windows into multiple categories according to predetermined settings. However, the applied method does not use the "Canny edge detection" algorithm in a Python computing environment, such as the Python computing environment of this invention.

[0021] literature" de Fácies em Perfis de The paper "AlgoritmoInteligente-Santos, Renata de Sena, UFOPA; Andrade, AndréJoséNeves, UFPA" discloses solutions for identifying lithology and lithofacies in well profiles using both Vsh-LK diagrams and generalized angle competition networks. However, the methods employed do not utilize the "Canny edge detection" algorithm in a Python computing environment, such as the Python computing environment described in this invention.

[0022] The document "Estudo de The paper "de Bordas em Imagens Usando Kernel-BrunaCavallero Martins, Matheus Fuhman Stigger, Wemerson Delcio Parreira" reveals the application of kernel functions in edge detection problems in images. Although this invention incorporates kernel filters, the document does not mention the configuration for using the "Canny edge detection" algorithm in a Python computational language environment, such as that described in this invention.

[0023] It can be seen that the prior art does not possess the unique features of this invention, which will be presented in detail below. No specific process has been identified for performing the following operation: extracting contour contrasts observed in cross-sectional views of images captured from rocks within a Python computational language environment, such as that described in this invention.

[0024] The effectiveness of the invention has been demonstrated in the identification of texture and structural patterns necessary for the correlation of rock cross-sections and lithofacies identification.

[0025] The effectiveness of this invention has also been demonstrated in artifact removal. Specifically, tool mark artifacts are removed without compromising the identification of rock structure and texture. Conversely, the removal of these artifacts improves the visualization of geological features. This application is a significant innovation because no other processing method currently exists that delivers such results with high quality.

[0026] The present invention also yields advantageous results in quantifying edge structures by depth. This is important for tomographic profiles, image profiles, and photographic evidence. Identified inhomogeneities may be related to texture variations and the presence of megapores and gigapores, thus aiding in lithofacies and productivity analysis. Understanding these edge contrasts can also indicate artifacts that have not been removed (such as, for example, fractures, cable traces, borehole marks, etc.). Quantifying edge contrasts, in terms of the occurrence of artifacts, can be a way for interpreters to quantify the quality of an image. Summary of the Invention

[0027] The development and improvement of computational scripts aimed at extracting contour contrasts from images contribute to the interpretation of image lithofacies. Texture and structural variations are highlighted, facilitating the identification of typical patterns.

[0028] In this invention, a tool was developed that highlights the inhomogeneity of rocks to allow for the initial purpose of identifying typical texture and structural patterns that may be associated with sedimentary rock facies.

[0029] This invention uses the "Canny edge detection" algorithm in a Python-based computational language environment to extract edge contour contrasts observed in captured rock images.

[0030] The improvements of this invention enable other objectives, such as: artifact removal, overlaying the original image with the detected edges, and quantifying the contrast of the identified edges by depth. Attached Figure Description

[0031] The invention will now be described in more detail with reference to the accompanying drawings, which illustrate examples of implementations of the invention in a illustrative and not limiting manner. The following figures are included in the drawings:

[0032] Figure 1 An algorithm for extracting edge contrast is shown, highlighting structural contours that can represent texture and facies variations. A) corresponds to an acoustic image profile; while B) corresponds to an image with detected edges.

[0033] Figure 2 The presence of megapores to gigapores, potentially associated with certain lithologies, is shown. Furthermore, artifact occurrence must be minimized during data acquisition and processing. The aim is to detect edge artifacts without compromising petrographic interpretation. A) Corresponds to an acoustic image profile. B) Corresponds to an image with detected edges. Note the detection of cable traces (artifacts);

[0034] Figure 3 The edge contrast (B, D) is shown to also be extracted from photographs of the lateral sample (A) and the petrographic slice (C), thus highlighting the textural differences in petrographic composition;

[0035] Figure 4 This demonstrates how to perform depth adjustments on fault scan profiles by using acoustic image profiles to correlate gamma-ray profiles (images) with natural gamma curves from cores (evidence), as well as to make subtle adjustments based on observed texture variations and structures.

[0036] Figure 5 This illustrates the repositioning of lateral samples using traces (indicated by arrows) observed in image profiles. A) Corresponds to an acoustic image profile. B) Corresponds to an electrical resistivity image profile—oil-based drilling fluid. C) Corresponds to an electrical resistivity image profile—water-based drilling fluid.

[0037] Figure 6 The image shows pixel classification, suppressing pixels that do not correspond to local maxima. If point A is a local maximum, it is considered to belong to an edge;

[0038] Figure 7 Local hysteresis is shown. Values ​​above the upper bound are considered true edges. Values ​​between the upper and lower bounds are considered edges if they are connected to true edges. Values ​​not connected are discarded. Values ​​below the lower bound are also discarded.

[0039] Figure 8 The product obtained from the script developed in this invention is shown. The original image (A) has edges (C) extracted from user-defined boundaries (B), which can overlap with the original image (D). The detected edge density (E) is also provided.

[0040] Figure 9The image shows stromatolites observed in an acoustic image profile (image lithofacies), where contour contrasts are highlighted through image processing using Python. A) Corresponds to a laminated stromatolite with pores having preferred paths along the lamellars.

[0041] B) Corresponding to stromatolites with cavities and pores that are characteristic of stromatolites and whose internal arrangement is significantly more disordered (more prevalent);

[0042] Figure 10 The diagram shows in-situ sediments scattered with redepositions. Finely textured lamellar (LMT) rocks, exhibiting early stratification or even no stratification (due to thickness below the tool's resolution). Spherulite (ESF), with high amplitude (closed) and fine to very fine granular texture. Stromatolites (STR), with low amplitude and porosity along preferred paths defined by landslides (elemental growth levels) and random cavitation (unless caused by increased inter-elemental porosity due to dissolution). Grainy limestone (GST), with low amplitude and granular texture. Generally, they are porous, except at more closed levels where the amplitude of the acoustic image profile is greater.

[0043] Figure 11 The redeposited sediments are shown. A) corresponds to a granular texture with low-amplitude pores. B) corresponds to rubble limestone (RUD).

[0044] C) corresponds to silicified pumice (FLT-sl) embedded with low-amplitude RUD. D) corresponds to layered grainy limestone (GST). Detailed Implementation

[0045] This invention relates to the development of computational scripts for processing image profiles, enabling the highlighting of texture and structural variations in rocks in any type of image. This information can be used in methods for determining lithofacies.

[0046] This invention uses the "Canny edge detection" algorithm in a Python-based computational language environment to extract edge contour contrasts observed in captured rock images.

[0047] Image processing that provides edge contour contrast is important because it highlights inhomogeneities that allow for the identification of texture and structural patterns. Figure 1 These changes indicate lithofacies variations and contribute to the interpretation of lithofacies in images.

[0048] Edge contrast can also represent mega-to-giga-hole or artifacts. Figure 2 ).

[0049] Image quality is crucial so that edge contrasts represent texture and structural variations, highlighting features presented in image profiles, tomographic images or evidence photographs, lateral samples, and thin slices. Figure 3 Therefore, the resulting products can contribute to the characterization of lithofacies in images and to the correlation of rock × profile diagrams.

[0050] Methods for performing rock × profile correlation

[0051] A method for determining lithofacies based on high-resolution image profiles includes the following steps:

[0052] a) Obtain computed tomography (CT) scans and generate CT profile images;

[0053] b) Data acquired in the well is referenced to the well depth, which is determined by measuring the length of the cable as it enters and exits the well. This is achieved by measuring the cable rotation close to the profiling unit. However, due to the tensile strength of the cable and the interaction between the tool and the wellbore roughness, the recorded depth may not be the true depth. The depth of the first profiling run is considered a reference because the cable deformation is minimal. Therefore, depth adjustments for the profile must be performed by correlating the natural gamma ray curve of the core measured in the laboratory with the reference gamma ray curve from the first profiling run.

[0054] c) Because the resolution of image profiles is greater than that of curved gamma rays, and because image profiles may contain artifacts, it may be necessary to perform fine-tuning of the image profile's depth after processing it. This fine-tuning is performed by examining the correlation between the texture and geological surface observed in the image profile and the tomographic profile or evidence. Figure 4 );

[0055] d) In addition to evidence, lateral sampling corresponds to another source of direct information about the rock. Lateral samples are taken from a pre-defined depth. Due to operational issues, samples are not always taken from the precisely predicted depth. The cavities left by lateral rock sampling are well identified in the image profile. Therefore, the lateral samples can be repositioned, and the depth from which the lateral samples were taken can be precisely indicated. Performing this repositioning of the lateral samples is very important;

[0056] e) In conjunction with the description of the thin sections, the previously described lithofacies are calibrated and (if necessary) the evidence is (re)described. From this stage onward, the use of the results provided by the scripts developed in this invention adds value to the method for determining the lithofacies of images;

[0057] f) Interpret the lithofacies in the acoustic image profile (image lithofacies);

[0058] g) Extrapolate the image lithofacies for depths of unproven and unsampled intervals (i.e., intervals between lateral samples).

[0059] If no available tomographic profile is available, the process begins with depth adjustment of the gamma-ray profile (image) and the core's natural gamma curve (evidence). If no evidence is available, the process begins with repositioning of the lateral sample, performed by adjusting the depth measured by the detector and by using traces observed on the image profile as a reference. Figure 5 The results of depth adjustment for lateral samples are not always reliable because, for the same depth, there may be several attempts to collect lateral samples (information indicated in the profiling report). The presence of mega- to giga-pores (such as voids), fractures, and / or cracks) increases uncertainty regarding the correct location of the sample. One method to increase the accuracy of repositioning lateral samples is to correlate the basic petrophysical data (porosity and permeability) with other profiles, such as magnetic resonance and neutron profiles, that show traces observed in the image profile. After repositioning the sample, the basic petrophysical data must also include its adjusted depth.

[0060] After calibrating the texture patterns of different lithofacies, the image lithofacies were extrapolated to the entire well, taking diagenesis into account. This calibration was performed by comparing the texture and structure between rock data and image profiles of different lithologies. Significant uncertainties were associated with areas where core samples were missing, areas where there were questions regarding the description of lateral samples, areas where diagenesis had eliminated the original rock identification, and areas where artifacts and image profile quality impaired the definition of image lithofacies. Notably, in addition to calibrating image profiles using rock data, integrated analysis with other profiles such as density, neutron, acoustic, caliper, photoelectric factor, resistivity, magnetic resonance, gamma-ray spectroscopy, and petrogeochemistry profiles contributes to the characterization of image lithofacies. The description of lateral samples provides greater reliability for lithofacies interpretation and extrapolation.

[0061] Extraction of contour contrast in images

[0062] The "Canny Edge Detection" algorithm available in the programming code library aims to extract observed contour contrasts in an image by performing the following operations:

[0063] 1) By applying 5 × 5. Use a Gaussian filter to reduce noise.

[0064] Once the Canny edge detection algorithm may introduce noise into the image, it is necessary to remove the noise by applying a Gaussian filter (1). The function of this filter is to smooth the image and then remove noise and details.

[0065]

[0066] Equation 1 represents a one-dimensional Gaussian function, in which G(x) corresponds to the Gaussian distribution of the values ​​of x, ρ corresponds to the standard deviation of the values ​​of x (such that ρ > 0), and x corresponds to a set of n values ​​(such that -∞). <x<∞)。

[0067] 2) Calculate the image intensity gradient.

[0068] For the smoothed image, its level G x Direction and vertical G y The intensity in the direction is further analyzed. The intensity gradient (edge ​​gradient) is calculated by applying a Sobel kernel filter to each pixel in the image, thus obtaining the direction of the maximum change from bright to dark and the amount of change in that direction. The gradient direction is always perpendicular to the edge and is rounded to one of four angles representing the horizontal, vertical, and two diagonal directions. Therefore, the image intensity gradient has an magnitude (2) and a direction (3).

[0069]

[0070] Equation 2 represents the magnitude G of the image intensity gradient, which is determined by the image intensity gradient at the horizontal level. x Direction and in vertical G y Strength calculation in the direction.

[0071]

[0072] Equation 3 represents the θ direction of the intensity gradient of the image.

[0073] 3) Eliminate pixels that do not correspond to true edges.

[0074] After obtaining the magnitude and direction of the gradient, a full image analysis is performed to remove pixels that do not constitute edges. Each pixel is evaluated to determine whether it constitutes a local maximum relative to its nearest neighbor along its gradient direction.

[0075] exist Figure 6 In the diagram, point A corresponds to an edge in the vertical direction, where the gradient direction is perpendicular to that edge. Points B and C lie in the gradient direction. Point A is compared with points B and C to verify whether point A corresponds to a local maximum. If it matches, point A is considered an edge and proceeds to the next stage. Otherwise, point A is suppressed (reset to zero). The resulting product is a binary image with the detected edges.

[0076] 4) Hysteresis limitation.

[0077] This step defines which previously selected edges are true edges and which are false positives. For this, it's necessary to insert two parameters that constitute a lower and upper bound. Pixels with intensity gradient values ​​greater than the upper bound are considered true edges, while those with values ​​less than the lower bound are not considered edges and are discarded. Pixels with intermediate values ​​are analyzed based on their connections to neighboring pixels. If a pixel is connected to a true edge pixel, it is considered an edge. If the pixel is not connected, it is discarded.

[0078] exist Figure 7 In the equation, part A of the edge is above the upper bound, so part A is considered a true edge. Although part C is between the lower and upper bounds, its connection to part A makes it a true edge. Part B, despite having a value close to that of C, has no connection to any true edges and is therefore discarded.

[0079] The "Canny edge detection" algorithm and its developed scripts are available in the OpenCV library.

[0080] The Canny edge detection algorithm exists in the OpenCV library, which is used for Python programming software. Through personalized parameterization within the Canny edge detection algorithm, we attempt to extract edge contour contrasts observed in images (i.e., image profiles, tomographic images or evidence photographs, lateral samples, thin sections). The aim is to highlight texture and structural variations to aid in interpreting the lithofacies of the images.

[0081] The developed script, in its latest version, allows for sequential processing of well images, provided specific conditions are met, such as format (*.PNG) and filename standardization. After importing, the well image can be analyzed based on its resolution in DPI, height and width in pixels and / or inches, and made available for application of the Sobe kernel filter. This filter is designed to smooth the image by eliminating noise. The generated image can be plotted, and chromaticity can be modified.

[0082] Subsequently, the user sets upper and lower limits for edge contrast detection to enhance texture and structural variations in the image. The same limits can be applied to all imported images, or the user can specify the most suitable values ​​for each case. After setting the upper and lower limits, the script can be activated only once to generate all products.

[0083] The detected edge images can be viewed individually or overlaid on the original image (overlapping images). To quantify the information, the detected edge density data is exported as a *.txt file, and the image of this data is exported as a *.PNG file.

[0084] Therefore, the products generated by the script are edge images, overlapping images, edge density images, and files in *.txt format. Figure 8 ).

[0085] result

[0086] The linked image profile × evidence is performed by adjusting for the depth of evidence regarding the natural gamma curve of the core and for variations in texture and structure observed in the acoustic image profile and / or electrical resistivity image profile. Calibration can be performed between the texture, structure, and amplitude observed in the image profile and fault scan images based on the lithofacies described in the evidence, which helps in determining the image lithofacies. When integrating this information with data from other profiles, petrographic slice descriptions, and laboratory petrophysical results, it may be necessary to reinterpret the lithofacies described in the database.

[0087] Stromatolites can have conical, dome-shaped, and / or lamellar shapes. In image cross-sections, stromatolites may exhibit landslides formed by small elements. Regions of low amplitude (porosity) can be observed along preferred paths defined by these landslides, corresponding to the level of element growth (…). Figure 9 A). On the other hand, cavitation pores appear randomly and have a significantly more disordered internal arrangement (A). Figure 9 B), unless they are the result of increased inter-elemental porosity due to dissolution treatment, and will show the geometry of stromatolites.

[0088] Spherulites have a fine granular pattern, which disappears in the presence of strong cementation and / or adhesion. Typically, in image cross-sections, spherulites tend to exhibit a laminated or more homogeneous structure when diagenesis is active and when the spherulites are below the resolution of the tool. Spherulites usually have very fine granular texture or fine grain. Distinguishing them from lamellar lithofacies is not always obvious.

[0089] Laminated rocks tend to have continuous landslides. Identification of laminated rocks largely depends on rock × profile correlation. Carbonate sludge, present as components of clay laminated rocks and spherulites, can eliminate pores and particles, making them difficult to distinguish. In these cases, the presence of siliceous detrital and carbonate materials exhibiting different impedance responses allows for relatively better contrast marking in laminated rocks. In acoustic image profiles, laminated rocks are characterized by embedded layers with amplitude contrast, early stratification, or even the absence of stratification (due to landslide thickness below tool resolution). In-situ sediments can appear scattered among each other or as redeposited lithofacies (…). Figure 10 ).

[0090] The redeposited lithofacies (conglomerate limestone, pumice, and grain limestone) are mainly characterized by low amplitude ( Figure 11 A) or high-amplitude granular texture. Low amplitude indicates coarse and porous walls. Layers and boulders with larger amplitudes correspond to dense layers (i.e., cemented or silicified layers). Conglomerate limestone with very coarse grains, as determined by grain size analysis, exhibits a distinct granular texture. Figure 11 B). Pumice with a thin-grained limestone matrix can exhibit a similar response to conglomerate limestone. Figure 11 C). Grainstone exhibits a more discrete grainy texture than conglomerate limestone. However, the distinction between these lithofacies is not always clear, especially when the profile resolution is low, or when diagenesis has taken effect, eliminating the original texture of the rock. Re-deposited sediments may have a laminated, layered, or massive structure. For example, the occurrence of layered grainstone is characterized by the presence of layers with amplitude contrasts due to variations in grain size. Layers comprising coarser and more porous grains exhibit low amplitude. Layers comprising finer grains and more closed (less porous) structures exhibit high amplitude. Figure 11 D).

[0091] Carbonate breccia is associated with exposure, weathering, and erosion. These processes tend to erase the forms that give rise to lithofacies. Silica intrusion via hydrothermal fluids also deforms pre-existing sediments and forms breccia, producing a high-amplitude response in acoustic image profiles. Breccia exhibits a disordered texture, which may or may not preserve the original stratification of the rock. When it is possible to define the protolith in the evidence, it can be incorporated into a classification (e.g., conglomerate limestone with breccia). When this is not possible, the breccia classification is used. Grainy limestone, dolomite, and siliceous rocks are also associated with diagenetic treatments that tend to erase past rock textures. Image lithofacies can be interpreted when there is calibration using the rock and other profiles, particularly photoelectric factor (PE) and petrogeochemical profiles.

Claims

1. An image processing method, characterized in that, The image processing method uses the Canny edge detection algorithm, which includes the following operations: a. Apply a 5×5 Gaussian filter to the image of the well to smooth the image; b. Calculate the image intensity gradient for each pixel in the smoothed image; c. Remove pixels that do not correspond to the edges of the image based on the image intensity gradient; as well as d. Determine which pixels correspond to true edges based on the upper and lower bounds of the image intensity gradient; The Canny edge detection algorithm further includes: extracting the edge contour contrast observed in the image, and quantifying the edge contour contrast by the depth of the well.

2. The method according to claim 1, characterized in that, The images include cross-sectional views, tomographic images or evidence photographs, lateral samples, and thin slices.

3. The method according to claim 2, characterized in that, The method highlights the texture and structural variations of the image.

4. The method according to claim 2, characterized in that, The upper and lower limits of the image intensity gradient are defined by the user.

5. The method according to claim 2, characterized in that, The method enables the removal of artifacts from the image.

6. The method according to claim 5, characterized in that, The artifacts include tool marks.

7. The method according to claim 2, characterized in that, Images with defined true edges can be viewed alone or overlaid on the original image.

8. The method according to claim 2, characterized in that, The method enables sequential processing of well images, analyzing the well images based on their resolution in DPI, and their height and width in pixels and / or inches.

9. The method according to claim 2, characterized in that, Edge density data Export the files, and the graphs of these data are... Export format.

10. A method for determining lithofacies based on high-resolution image profiles, the method using the method defined in claim 1 or 2, characterized in that, The method includes the following steps: a) Obtain computed tomography (CT) scans and generate CT profile images; b) Adjust the depth of the profile by the correlation between the natural gamma curve of the core measured in the laboratory and the reference gamma curve of the first analysis run; c) Fine adjustments to the depth are made using the correlation between the texture and geological surface observed in the image profile and the fault scan profile or evidence; d) Reposition the lateral samples; e) To calibrate the previously described lithofacies and to describe or re-describe the evidence in conjunction with the description of thin sections; f) Using the results provided by the method as defined in claim 1 in the method for determining the lithofacies image; g) Interpret the lithofacies of the image in the acoustic image profile; h) Extrapolate the image lithofacies for the depth of the unproven and unsampled intervals, i.e., the intervals between lateral samples.

11. The method for determining lithofacies based on high-resolution image profiles according to claim 10, characterized in that, The method, without step a, begins with depth adjustment of the gamma-ray profile and the natural gamma rays in the core.

12. The method for determining lithofacies based on high-resolution image profiles according to claim 10, characterized in that, The repositioning of the lateral sample is performed by adjusting the depths measured by the drilling machine and by examining the traces observed in the image profile.

13. The method for determining lithofacies based on high-resolution image profiles according to claim 10, characterized in that, By correlating the basic rock physics data with other cross-sections that show traces observed in the image cross-section, the accuracy of the lateral sample repositioning can be increased.

14. The method for determining lithofacies based on high-resolution image profiles according to claim 13, characterized in that, The basic rock physics data includes porosity and permeability.

15. The method for determining lithofacies based on high-resolution image profiles according to claim 13 or 14, characterized in that, The other cross-sectional views include magnetic resonance and neutron cross-sectional views.

16. The method for determining lithofacies based on high-resolution image profiles according to claim 10, characterized in that, After calibrating the texture patterns of different lithofacies, the image lithofacies can be extrapolated to the entire well.

17. The method for determining lithofacies based on high-resolution image profiles according to claim 10, characterized in that, In addition to calibrating the image profiles using rock data, integrated analysis with other profiles helps to characterize the lithofacies of the images.

18. The method for determining lithofacies based on high-resolution image profiles according to claim 13 or 17, characterized in that, The other profiles can be density, neutron, acoustic, caliper, photoelectric factor, resistivity, magnetic resonance, gamma ray spectroscopy, and rock geochemical profiles.

19. The method for determining lithofacies based on high-resolution image profiles according to claim 10, characterized in that, In operation b, the Sobe kernel filter is applied for each image pixel.

Images