Evaluation of stereoscopic photometric images of rock debris particles

Abstract

A method for generating a segmented image of a cuttings particle includes obtaining and preparing a cuttings particle for imaging, and placing the prepared cuttings particle in front of a digital camera. At least three digital images of the cuttings particle are obtained at respective non-coplanar illumination angles, which are then combined to generate a photometric stereo image of the cuttings particle. A segmented image can be generated from the photometric stereo image.

Description

[0001] Cross-references to related applications

[0002] This application claims the benefit and priority of U.S. non-provisional application No. 19 / 034,859, filed January 23, 2025, the entire contents of which are incorporated herein by reference. Background Technology

[0003] Rock cuttings are generated during drilling operations used for oil and gas exploration and recovery, geothermal and scientific exploration. It should be understood that rock cuttings are abundant in both volume and quantity, and can provide one of the lowest-cost and richest data sources for understanding and characterizing the properties of subsurface rocks and formations. Rock cuttings have long been assessed on the surface to generate a detailed record of their properties. In addition, digital images of rock cuttings are sometimes acquired and then analyzed by off-site geologists. This assessment is both time-consuming and expensive.

[0004] While the aforementioned practices for assessing rock fragments are commercially available, further automation is needed. In recent years, methods have been published for automatically assessing digital images of rock fragments using digital image processing and artificial intelligence (AI) techniques to automatically segment individual fragments, classify rock fragment lithology, and estimate formation porosity. While these new methods are promising, there is room for further improvement. Attached Figure Description

[0005] To gain a more complete understanding of the disclosed subject matter and its advantages, reference is now made to the following description in conjunction with the accompanying drawings:

[0006] Figure 1 An example drilling rig, including an example system for evaluating rock cuttings, is depicted.

[0007] Figure 2 A flowchart depicts an example method for generating segmented images of rock debris particles.

[0008] Figure 3 An exemplary digital image depicting a densely arranged array of rock cuttings set on a tray.

[0009] Figure 4 A schematic diagram depicts the process of illuminating rock fragments from three non-coplanar light directions to obtain corresponding digital and photometric stereoscopic images.

[0010] Figure 5 A flowchart depicts an example method for segmenting individual rock cutting particles in an image of densely arranged rock cutting particles.

[0011] Figure 6 An example implementation is described in which stationary rock debris particles are illuminated from four different light source orientations during image acquisition.

[0012] Figure 7A An example system for generating segmented images of drilling cuttings particles is described.

[0013] Figure 7B Another example system for generating segmented images of drilling cuttings particles is depicted. Detailed Implementation

[0014] Embodiments of this disclosure include systems and methods for evaluating cuttings particles generated during underground drilling operations. In one example embodiment, a method for generating a segmented image of cuttings particles includes acquiring and preparing the cuttings particles for imaging, and placing the prepared cuttings particles in front of a digital camera. At least three digital images of the cuttings particles are acquired at corresponding non-coplanar illumination angles and then combined to generate a photometric stereo image of the cuttings particles. Segmented images identifying individual cuttings particles can be generated from the photometric stereo image. Features of the subsurface strata can then be estimated from the segmented images.

[0015] Figure 1 An example drilling rig 20 is depicted, which includes a system 100 for evaluating cuttings particles removed from circulating drilling fluid on the rig. The drilling rig 20 may be positioned above subsurface formations (not shown). The drilling rig 20 may include, for example, a derrick and lifting equipment (also not shown) for raising and lowering a drill string 30, which, as shown, extends into a wellbore 40 and includes, for example, a drill bit 32 and one or more downhole measurement tools 38 (e.g., logging-while-drilling tools or measurement-while-drilling tools) in a bottom-hole assembly (BHA) above the drill bit 32. Suitable drilling systems (e.g., including drilling, steerable, logging, and other downhole tools) are well known in the art.

[0016] The drilling rig 20 also includes a surface system 50 for controlling the flow of drilling fluids used on the rig (e.g., for drilling the wellbore 40). In the illustrated example rig, drilling fluid 35 is pumped downhole, for example, via a conventional mud pump 57 (as depicted at 62). Drilling fluid 35 can be pumped to the drill string 30, for example, through a riser 58 and mud hose 59. Drilling fluid 35 typically flows out of the drill string 30 at or near the drill bit 32, creating an upward mud flow 64 through the wellbore annulus 42 (the annular space between the drill string and the wellbore wall). Drilling fluid 35 then flows through a return pipe 52 to a mud pit system 56, where it can be recirculated. It should be understood that the terms drilling fluid and mud are used synonymously herein.

[0017] The circulating drilling fluid 35 is designed to perform a number of functions during drilling operations, one of which is to transport drilling cuttings 45 to the surface (in upward flow 64). Drilling cuttings 45 are typically removed from the return mud via a mud vibrating screen 55 (or other similar solids control device) in the return pipe (e.g., upstream of the mud pit 56). Formation gases released during drilling may also be carried to the surface in the circulating drilling fluid. These gases are typically removed from the fluid, for example, by a degasser or gas trap 54 located in or near a manifold 53, which, in the exemplary description, is located upstream of the mud vibrating screen 55. Drilling cuttings 45 can be evaluated to characterize the subsurface formation and / or estimate its various properties, as described in more detail below.

[0018] Drilling rig 20 may include system 200 configured to capture and evaluate digital images of drilling cuttings, as described in more detail below. System 200 may be deployed at the drilling rig site (e.g., in field laboratory 80) or off-site. However, the disclosed embodiments are not limited in this respect. System 200 may include computer hardware and software configured to automatically or semi-automatically evaluate the images of drilling cuttings. To perform these functions, the hardware may include one or more processors (e.g., microprocessors) that can be connected to one or more data storage devices (e.g., hard disk drives or solid-state storage). As known to those skilled in the art, the processor may further be connected to a network, for example, to receive images from a networked camera system (not shown) or another computer system. Of course, it should be understood that the disclosed embodiments are not limited to the use or configuration of any particular computer hardware and / or software.

[0019] Although Figure 1 A land-based drilling rig 20 is depicted, but it should be understood that the disclosed embodiments are equally applicable to land-based or offshore drilling rigs. As is known to those skilled in the art, offshore drilling platforms typically comprise a platform deployed atop a riser extending from the seabed to the water surface. The drill string extends downward from the platform, through the riser, and into the wellbore via a blowout preventer (BOP) located on the seabed. The disclosed embodiments are not limited to these aspects.

[0020] In recent years, methods have been published for automatically evaluating digital images of rock fragments using digital image processing and artificial intelligence (AI) technologies to automatically segment individual rock fragments and classify rock fragment lithology. While these new methods are promising, image segmentation of overlapping or piled objects (such as rock fragments of small rocks) presents inherent difficulties due to the limitations of 2D imaging, which cannot fully capture the 3D information of the object.

[0021] For example, rock debris grains have complex and irregular shapes, leading to occlusion where one object may hide or partially cover another. Distinguishing between overlapping or contacting objects in a 2D image is challenging, making accurate segmentation difficult. Furthermore, the loss of depth or height information about objects in 2D images makes it difficult to determine which parts of the grain are closer to the camera and which are farther away. This can result in segmentation ambiguity. Shadows can add further complexity to the segmentation task, as distinguishing between actual object boundaries and shadow edges can be difficult in 2D. Moreover, rock debris grains often have similar textures or color patterns, making it challenging to accurately distinguish them using traditional 2D image processing techniques. Further, the arrangement of rock debris grains on a tray or in a grain pile can vary significantly with different shapes, densities, and grain orientations, making it difficult to create a segmentation solution sized to fit all needs.

[0022] Another challenge is that complex segmentation tasks require large datasets of annotated images to train deep learning algorithms. Data annotation for deep learning segmentation incurs labor costs for manual labeling, tooling and infrastructure costs, and additional effort for quality control. Data training costs involve investing in powerful hardware, longer training times, and fine-tuning hyperparameters. Balancing these costs with the potential benefits is crucial to ensuring the cost-effectiveness and successful deployment of deep learning solutions.

[0023] Furthermore, deep learning algorithms often need to be retrained on specific use cases to achieve optimal results. This retraining is a complex and time-consuming process, requiring access to relevant data, expertise in model tuning, and an understanding of the difficulties of the specific use case. For effective retraining, sufficient and diverse data representing the specific use case is required. If data usage is limited to a specific project or geographic location, acquiring and preparing such data can be challenging or even prohibitive. Retraining can also be computationally intensive, requiring powerful hardware and sufficient training time. After retraining, thorough validation is essential to ensure that the model's performance truly improves under the target use case without adversely affecting performance in other scenarios. This validation process can also be time-consuming and resource-intensive. While retraining can (and often does) lead to improved results for a specific use case, it can negatively impact the model's performance in other tasks or scenarios. For these and other reasons, there is a need for improved methods for acquiring rock debris particles and deep learning segmentation methods in industry.

[0024] Figure 2A flowchart depicts an example method 100 for generating segmented images of rock cuttings (e.g., piled-up rock cuttings). At 102, rock cuttings are acquired and prepared for imaging. At 104, the rock cuttings are placed in front of a digital camera. In an exemplary embodiment, the rock cuttings may be in contact with each other. In other exemplary embodiments, at least a portion of the rock cuttings may pile up on top of other rock cuttings. At 106, at least three (e.g., at least four) digital images of the rock cuttings are acquired at corresponding non-coplanar illumination angles. At 108, the acquired digital images are combined to generate a stereoscopic image. At 110, individual particles within the piled-up rock cuttings are identified in the stereoscopic image to generate the segmented image. Optionally, the segmented images may be further processed to estimate one or more formation characteristics, such as formation lithology and / or formation porosity, as disclosed in commonly assigned U.S. Patent Publication 2023 / 0220770 and WIPO Publication WO2024 / 020523.

[0025] Compared to deep learning or artificial intelligence methods used for object segmentation, the disclosed segmentation method can offer several advantages. For example, the disclosed embodiments can provide depth information, leading to better separation of particles in an image and enabling accurate 3D reconstruction. Furthermore, this 3D reconstruction can further enhance texture variations both between and within particles. Moreover, the disclosed embodiments do not heavily rely on training with annotated data, thus significantly reducing cost and time. In example embodiments, real-time segmentation can be achieved with minimal computation time.

[0026] Rock cuttings can be obtained and prepared at 102 of method 100, for example, by drilling a well into or through the subsurface formation of interest, for example, using the methods described above. Figure 1 The example drilling rig 20 described. During drilling, rock cuttings are transported to the surface in the upward-flowing drilling fluid (e.g., as...). Figure 1 (As shown). Cutting particles can be collected, for example, using a mud vibrating screen or other solids separation / control equipment on the drilling rig. The collected particles are often contaminated with oil-based mud (OBM) or water-based mud (WBM), which may require further preparation.

[0027] To remove this contamination, the cuttings particles can be cleaned, for example, in a cleaning solution comprising a suitable solvent (e.g., acetone, ethanol, isopropanol, or water) and a detergent or surfactant. The cleaning solution is designed to effectively dissolve or soften drilling fluid or other contaminants adhering to the particle surface. Cleaning may further include physical abrasion or agitation to facilitate contaminant removal. Such processes may include, for example, mechanical brushing or scraping, liquid jetting, compressed air, and / or ultrasonic agitation. After cleaning, the cuttings particles can be rinsed with clean water to remove any residual cleaning solution, detergent, and / or solvent. The particles can then be dried and placed on a tray for imaging.

[0028] Figure 3 An exemplary digital image depicting rock cutting particles 95 set on a tray (e.g., placed on...) Figure 2 (In front of the digital camera at position 104). Note that the rock fragments are in a dense configuration, where each fragment is in contact with each other, sometimes partially overlapping with other fragments, and may even be stacked on top of each other. This contrasts with a sparse configuration in which no fragments are in contact with or overlapping each other (such a configuration is not shown). It should be understood that the disclosed embodiments can be advantageously used to evaluate digital images of dense grain configurations, for example... Figure 3 As shown.

[0029] Understandably, one of the main challenges in imaging densely packed rock debris grains is generating images with sufficient contrast to effectively distinguish individual grains while preserving intra-grain contrast to assess texture and textural differences. Figure 3 As shown, individual particles can exhibit variations in height, such as rising from or sinking into the image plane, and possess their own unique texture and color. There is a need to improve and enhance image contrast for particle segmentation and subsequent characterization, especially for densely packed particles.

[0030] While the disclosed embodiments can be advantageously used to generate segmented images of accumulated rock debris particles, it should be understood that this disclosure is not limited thereto. Furthermore, it should be understood that in a sparser configuration, placing the rock debris particles on a tray with a high-contrast (vibrant) background color to enhance subsequent particle identification and segmentation in the acquired image may be advantageous. Example colors include pure magenta (e.g., RGB values ​​of 255, 0, 255), pure blue (e.g., RGB values ​​of 0, 0, 255), pure green (e.g., RGB values ​​of 0, 255, 0), etc. Typically, such colors do not exist in nature and can therefore enhance the disclosed segmentation methods; however, the disclosed embodiments are explicitly not limited in this respect.

[0031] Now go to Figure 4The generation of stereo images at points 106 and 108 is described in more detail. Photometric stereo combines images obtained with varying directional illumination to analyze shadows and reflections, thereby enhancing image contrast. Figure 4 In the schematic diagram, rock fragments 95 are illuminated using at least three non-coplanar light directions 121, 122, and 123 to obtain at least three corresponding digital images. Occlusion and shadows on the object surface vary with lighting conditions, making it difficult to capture all surface details in a single image. By observing the object from the same viewpoint under different lighting conditions, photometric stereo analysis is performed to assess (calculate) the surface normal vector at selected pixels in the image (e.g., at each pixel). And albedo and reflectance (α (alpha)).

[0032] This analysis assumes a fixed camera position and constant camera settings during image capture (the only variation being the illumination direction, as shown in the figure). The resulting images are combined to create a synthetic image, allowing for local estimation of surface orientation and curvature. This analysis can be based on Lambertian reflectance, which assumes an ideal matte surface with uniform radiation in all directions, but can be further extended to accommodate non-Lambertian reflectance models such as the Phong, Torrance-Sparrow, and Ward models, broadening the potential of this technique.

[0033] For example, photometric stereo operations can utilize material reflectivity and object surface curvature to calculate enhanced images based on a Lambertian (matte, diffuse) surface. The diffuse intensity (I) is proportional to the angle between the incident light direction (L) and the object's surface normal (n), driven by the albedo-reflectance (alpha), which follows the Lambertian cosine law. Albedo represents the fraction of incident sunlight reflected by the surface. The surface normal (n) and albedo-reflectance (alpha) can be determined from images acquired from at least three non-coplanar light directions, assuming distant and parallel light rays, and the known light direction (L) is predetermined within the illumination setup, where ( )).

[0034] The calculated surface normal vectors can advantageously reveal fundamental information about the surface of rock fragments, even disclosing surface irregularities such as scratches, debris, indentations, and / or etching patterns, despite the preference for smooth surfaces. Furthermore, enhanced contrast enables improved grain segmentation. While a minimum of three images are typically required to determine the normals, practical applications can utilize more images (such as four or more, or even five or more) to reduce inherent imaging noise and improve image accuracy. Redundancy from multiple images can yield better analytical results, typically requiring a minimum of four images.

[0035] Now go to Figure 5This illustrates the use of individual rock fragments in a dense configuration (e.g., in...). Figure 2 The flowchart of example method 150 (at 110 of method 100) is shown. Method 150 advantageously integrates (or combines) surface normal and shading information to achieve more robust segmentation of dense grain configurations. The disclosed embodiments advantageously utilize geometric and shading cues to achieve more accurate and robust object segmentation, and may be particularly advantageous when rock fragments have complex shapes and textures and / or dense configurations (e.g., stacked on top of each other).

[0036] Continue to refer to Figure 5 At location 152, at the corresponding non-coplanar illumination angle, acquire at least three (e.g., at least four) digital images of the rock fragments (e.g., as described above regarding...). Figure 2 At 154, the acquired digital images are combined to calculate the surface normal vector. Initial segmentation can be determined at 156 based on the calculated surface normal vector. At 158, shadows are detected and extracted from the image. This shadow extraction can utilize any suitable algorithm, such as thresholding, gradient-based methods, or machine learning. At 160, the shadow direction is estimated, for example, by analyzing the shadow length and orientation relative to the light source direction. Shadow-based segmentation can be performed at 162, for example, using the shadow information obtained at 160 to distinguish individual particles from each other. The resulting shadows act as or provide additional boundaries between particles and / or background between different particles. By considering shadow cues, separated particles can be identified more accurately. The surface normal segmentation obtained at 156 and the shadow-based segmentation obtained at 162 can be combined or integrated at 164 to obtain improved segmentation. Optional edge detection techniques can be applied to the integrated segmentation at 166 to further refine particle boundaries. Furthermore, optional region growing techniques can be applied to the integrated segmentation at 168 to obtain improved segmentation. For example, pixels with similar normal vectors and shadow properties can be grouped together to segment meaningful granular regions. Furthermore, post-processing techniques such as morphological operations, noise reduction, and object appearance / splitting can be optionally applied to improve the final segmentation results.

[0037] It should be understood that in many imaging applications, such as when selecting the size of a lighting system for a scene, the assumption of using distant light with parallel illumination rays is reasonable. A variety of suitable lighting tools, such as segmented strips and ring lights provided by companies like Advanced Illumination (Rochester, Vermont), CCS (Boston, Massachusetts), or SmartVision Lights (Muskgan, Michigan), are readily available and can be utilized. Such specialized lights greatly facilitate integration and setup, especially for machine vision software providers offering photometric stereo tools, such as Matrox Imaging (Montreal, QC, Canada).

[0038] It will be further understood that when the direction and intensity of illumination are known, the photometric stereo can be efficiently solved as a linear system. The illumination position can be determined based on the geometry of the illumination setup, or calibrated from an image using a specular sphere. However, a more challenging problem arises when the illumination details are unknown: uncalibrated photometric stereos. While solutions exist for calculating photometric stereos in these cases, it should be noted that uncalibrated methods are more sensitive to acquisition conditions and may suffer from poor repeatability, highlighting the importance of having accurate illumination information to obtain reliable results.

[0039] In some example embodiments, the rock fragments remain stationary during image acquisition, and estimated surface normal vectors and albedo (alpha) results are calculated. The albedo results provide an estimated percentage of the surface reflectance of the rock fragments, revealing variations in surface reflectance from glossy to dull. These variations in diffuse reflectance can indicate differences in material properties, resulting in enhanced visual contrast. This contrast enhancement is beneficial for segmentation and subsequent image analysis. Dramatic changes in the surface normal may indicate the presence of defects such as cracks, scratches, or dents. Beyond identifying defects, the surface normal vectors can be further processed to estimate local surface curvature. Analyzing these curvature results can be more intuitive than analyzing the entire normal vector field because they highlight local variations in the normal direction, such as protruding marks or depressions on an otherwise flat surface.

[0040] In other exemplary embodiments, rock fragments may be moved, for example, on a conveyor during image acquisition. To accommodate object movement, additional leading and trailing images may be captured under full illumination. These additional images can be used to determine the object's displacement, thereby allowing image realignment to compensate for the object's position at different times. The image between the leading and trailing images (photometric stereo source images captured with directional illumination) is then translated to ensure that the object's position is consistent in each image. High-speed cameras can be advantageously used to minimize perspective distortion and parallax errors caused by small degrees of displacement during object movement.

[0041] In some embodiments, using four different lights may be challenging in certain settings or for certain objects due to physical constraints or practical limitations. A possible alternative to using multiple fixed lights is to use a rotary table to rotate the sample (a tray of rock cuttings). This technique involves rotating the tray of rock cuttings instead of positioning multiple fixed lights around it. For example, a single light source positioned at a fixed angle relative to the object can be used to illuminate the object from a consistent direction throughout the rotation. The particles can be placed on a turntable (or other rotating device) that allows it to rotate smoothly and accurately about a single axis. Images can then be captured during rotation at predetermined angular orientations (e.g., at 60-degree or 90-degree intervals). The number of images taken may depend on the required accuracy and level of detail for reconstruction; more images can provide better 3D shape reconstruction, but at the cost of increased time and processing requirements.

[0042] Figure 6 An exemplary implementation in which rock cuttings remain stationary during image acquisition is depicted. In this example, four different images 131, 132, 133, and 134 are generated at corresponding different light source orientations (referred to herein as east, north, south, and west). Note that the rock cuttings are piled up and densely configured, as described above regarding... Figure 3 Furthermore, the particles have a wide range of sizes and shapes. Figure 6 The shading analysis of each image at points 135, 136, 137, and 138 is further depicted. The final photometric stereo image is depicted at point 140. As depicted, the final photometric stereo image 140 has sharper contrast and boundary delineation than any individual image (shown side-by-side with image 131). This contrast and boundary delineation are particularly well observed at points 140i and 131i. This contrast and grain delineation can advantageously improve the efficiency and accuracy of grain segmentation algorithms, for example, as described above.

[0043] Figure 7 depicts an example system 200 for generating segmented images of rock cuttings. It should be understood that the disclosed embodiments are not limited to any particular system configuration. As described above, the exemplary system 200 may include a stage or tray 220 configured to receive rock cuttings. The size and shape of the tray may be designed (in coordination with the camera lens system) to accommodate a sufficient quantity of rock cuttings and fill the field of view in the image acquisition apparatus. System 200 may include substantially any suitable tray, for example, including plastic or metal trays as described above; however, in a preferred embodiment, it includes a rotatable stage as described above and indicated by 221.

[0044] System 200 may further include at least one light source 215, for example, a white light source configured to illuminate the sample holder 220 (and rock fragments placed on the tray). The light source 215 may include, for example, a light-emitting diode (LED), a diode array, or other suitable light source capable of illuminating the rock fragment particles on the holder 220. System 200 may also include a camera 210 and a corresponding lens 211 deployed above the sample holder 220, for example, mounted on or within a separator disposed between the upper and lower chambers of system 220. The camera 210 may advantageously include a high-resolution color camera, for example, including a 10 or 20 (or more) megapixel image sensor. Essentially any suitable lens 211 can be used. The lens 211 may be configured to provide a sharp (focused) image to the image sensor, for example, including a 25 mm lens. Alternatively, the lens 211 may include a variable zoom lens.

[0045] System 200 may also include a controller 230, such as a computer board (or motherboard), configured to control the operation of camera 210, lens 211, lamp 215, and / or rotating sample holder 220. Controller 200 may include one or more processors (e.g., microprocessors) that can be connected to one or more data storage devices (e.g., hard disk drives or solid-state storage). The controller may be configured to network (e.g., communicate) with external devices (e.g., external computer system 240), for example, via a hardwired or wireless connection. For example, controller 230 may be configured to upload acquired images to computer system 240 for use such as... Figure 2 and 5 The processing described herein. In such an embodiment, controller 230 and / or computer system 240 may include processor-executable instructions stored in memory to perform the above-described processing. Figure 2 and Figure 5 The selection step in the described method steps.

[0046] Figure 7BThis is another example system 250 for generating segmented images of rock cuttings. System 250 is similar to system 200 in that it includes a stage or tray 275 configured to receive rock cuttings, and a camera 210 and a corresponding lens 211 deployed above the sample holder 275 and configured to acquire digital images of the rock cuttings in the tray. System 250 may also include a controller 230 and a computer system 240, as described above regarding... Figure 7A As described. In Figure 7B In this system 250, first, second, third, and fourth lamps 265a, 265b, 265c, and 265d (e.g., strip lights) are deployed around the sample holder and configured to illuminate the sample holder (e.g., from four different directions spaced apart at approximately 90 degrees). This system 250 can be advantageously used with or without a rotatable tray or stage (a non-rotating stage 275 is depicted). In some advantageous embodiments, lamps 265a, 265b, 265c, and 265d can be deployed on platform 270 (or a divider) and directed to illuminate rock fragments on the non-rotating stage 275. Lamps 265a, 265b, 265c, and 265d can alternatively be deployed on the sidewalls of the chamber. The disclosed embodiments are, of course, not limited in this respect. Furthermore, it should be understood that the disclosed embodiments are not limited to the use of a rotatable stage (e.g., strip lights). Figure 7A (as shown) or use multiple lights (such as) Figure 7B As shown in the figure, the system only needs to be able to acquire three or more images in the corresponding lighting direction.

[0047] Continue to refer to Figure 7A and 7B It should be understood that systems 200 and 250 can be used to acquire multiple necessary digital images, and then process or evaluate these images to obtain segmented images. For example, in Figure 7A In this system, system 200 can be used to rotate the rotatable stage 220 to at least three different angular orientations (e.g., four different angular orientations) and acquire different digital images of rock cuttings at each corresponding angular orientation. Figure 7B In this system, system 250 can be used to selectively and individually illuminate each light source (e.g., lamps 265a, 265b, 265c, 265d) and acquire different digital images of the rock debris particles corresponding to each individual illumination.

[0048] It should be understood that this disclosure includes many embodiments. These embodiments include, but are not limited to, the following embodiments.

[0049] In a first embodiment, a method for generating a segmented image of rock cutting particles includes: acquiring and preparing rock cutting particles for imaging; placing the prepared rock cutting particles in front of a digital camera; acquiring at least three digital images of the rock cutting particles at corresponding non-coplanar illumination angles; combining the at least three digital images to generate a photometric stereo image of the rock cutting particles; and generating a segmented image that identifies individual rock cutting particles within the rock cutting particles.

[0050] The second embodiment may include the first embodiment, wherein the acquisition and preparation further includes drilling a subsurface wellbore; collecting the cuttings particles from circulating drilling fluid; washing the collected cuttings particles; and rinsing the washed cuttings particles.

[0051] The third embodiment may include any one of the first to second embodiments, wherein the placement includes placing the prepared rock cutting particles on a rotatable stage in front of the digital camera; and the acquisition includes rotating the rotatable stage to at least three different angular orientations, and acquiring a digital image of the rock cutting particles at each of the at least three different angular orientations.

[0052] The fourth embodiment may include any one of the first to third embodiments, wherein the placement includes placing the prepared rock cutting particles on a platform in front of the digital camera and at least three light sources configured to illuminate the platform at correspondingly different angular orientations; and the acquisition includes selectively and individually illuminating each of the at least three light sources and acquiring a digital image of the rock cutting particles corresponding to each of the individual illuminations of the at least three light sources.

[0053] The fifth embodiment may include any one of the first to fourth embodiments, wherein the combination includes calculating the surface normal vector at selected pixels in the photometric stereo image. And albedo and reflectance (alpha); and generating segmented images also includes identifying individual rock fragments within the rock fragments based on the calculated surface normal vector.

[0054] The sixth embodiment may include any one of the first to fifth embodiments, wherein generating the segmented image further includes detecting and extracting shadows from at least three digital images; estimating the orientation of a selected shadow among the extracted shadows; and identifying individual rock fragments in the rock fragments based on the estimated shadow orientation.

[0055] The seventh embodiment may include any one of the first to sixth embodiments, wherein generating the segmented image further includes generating a first segmented image in which individual rock fragments are identified based on surface normal vectors calculated in the photometric stereo image; and generating a second segmented image in which individual rock fragments are identified based on estimated directions of shadows extracted from the at least three digital images.

[0056] The eighth embodiment may include the seventh embodiment, wherein generating the segmented image further includes combining the first segmented image and the second segmented image to obtain the third segmented image.

[0057] The ninth embodiment may include the eighth embodiment, wherein generating the segmented image further includes applying edge detection technology or region growing technology to the third segmented image.

[0058] The tenth embodiment may include any one of the first to ninth embodiments, wherein the acquisition, the combination, and the generation are performed automatically.

[0059] The eleventh embodiment may include any one of the first to tenth embodiments, and it further includes estimating the features of the subsurface strata from the segmented image.

[0060] In a twelfth embodiment, a system for generating segmented images of rock fragments includes: a sample holder configured to receive rock fragments; a digital camera positioned and configured to record digital images of the rock fragments on the sample holder; at least one light source configured to illuminate the sample holder; and a controller configured to (i) cause the digital camera to acquire at least three digital images of the rock fragments at corresponding non-coplanar illumination angles; (ii) combine the at least three digital images to generate a photometric stereo image of the rock fragments; and (iii) generate a segmented image identifying individual rock fragments within the rock fragments.

[0061] The thirteenth embodiment may include the twelfth embodiment, wherein the controller is further configured to rotate the sample holder to at least three different angular orientations corresponding to the non-coplanar illumination angle when at least three digital images of the rock cuttings are acquired.

[0062] The fourteenth embodiment may include any of the twelfth to thirteenth embodiments, wherein combining at least three digital images further includes calculating the surface normal vector at selected pixels in the photometric stereo image. And albedo and reflectance (alpha).

[0063] The fifteenth embodiment may include the fourteenth embodiment, wherein generating the segmented image further includes: generating a first segmented image, in which individual rock fragments are identified based on surface normal vectors calculated in a photometric stereo image; generating a second segmented image, in which individual rock fragments are identified based on estimated directions of shadows extracted from the at least three digital images; and combining the first segmented image and the second segmented image to obtain a third segmented image.

[0064] In a sixteenth embodiment, a method for generating a segmented image of rock debris particles includes: acquiring and preparing rock debris particles for imaging; placing the prepared rock debris particles in front of a digital camera; acquiring at least three digital images of the rock debris particles at corresponding non-coplanar illumination angles; combining the at least three digital images to generate a photometric stereo image of the rock debris particles; generating a first segmented image, in which individual rock debris particles are identified based on a calculated surface normal vector in the photometric stereo image; generating a second segmented image, in which individual rock debris particles are identified based on an estimated direction of shadow extracted from the at least three digital images; and combining the first segmented image and the second segmented image to obtain a third segmented image.

[0065] The seventeenth embodiment may include the sixteenth embodiment, wherein the acquisition and preparation further includes drilling a subsurface wellbore; collecting the cuttings particles from circulating drilling fluid; washing the collected cuttings particles; and rinsing the washed cuttings particles.

[0066] The eighteenth embodiment may include any one of the sixteenth to seventeenth embodiments, wherein the placement includes placing the prepared rock cutting particles on a rotatable stage in front of the digital camera; and the acquisition includes rotating the rotatable stage to at least three different angular orientations, and acquiring a digital image of the rock cutting particles at each of the at least three different angular orientations.

[0067] The nineteenth embodiment may include any one of the sixteenth to eighteenth embodiments, wherein the placement includes placing the prepared rock cutting particles on a stage in front of the digital camera and at least three light sources configured to illuminate the stage at correspondingly different angular orientations; and the acquisition includes selectively and individually illuminating each of the at least three light sources and acquiring a digital image of the rock cutting particles corresponding to each of the individual illuminations of the at least three light sources.

[0068] The twentieth embodiment may include any one of the sixteenth to nineteenth embodiments, and further includes applying edge detection technology or region growing technology to the third segmented image.

[0069] The twenty-first embodiment may include any one of the sixteenth to twentyth embodiments, and it further includes estimating the features of the subsurface strata from the third segmented image.

[0070] Although photometric stereoscopic images of rock fragments have been described in detail, it should be understood that various changes, substitutions and alterations may be made herein without departing from the spirit and scope of this disclosure as defined by the appended claims.

Claims

1. A method for generating a segmented image of rock debris particles, the method comprising: Obtain and prepare rock cuttings for imaging; The prepared rock cutting particles are placed in front of a digital camera; At least three digital images of the rock fragments were acquired at the corresponding non-coplanar illumination angles; The at least three digital images are combined to generate a photometric stereoscopic image of the rock fragments; and A segmented image is generated to identify each individual rock fragment in the rock fragments.

2. The method according to claim 1, wherein the acquisition and preparation further comprises: Drill an underground well; The rock cuttings are collected from the circulating drilling fluid; Wash the collected rock fragments; and Rinse the washed rock fragments.

3. The method according to claim 1, wherein: The placement includes placing the prepared rock cuttings on a stage in front of the digital camera and at least three light sources, the at least three light sources being configured to illuminate the stage at correspondingly different angular orientations; The acquisition includes selectively and individually illuminating each of the at least three light sources, and acquiring a digital image of the rock debris particles corresponding to each of the individual illuminations from the at least three light sources; and The combination includes calculating the surface normal vector at selected pixels in the photometric stereo image. And albedo and reflectance.

4. The method according to claim 1, wherein generating the segmented image further comprises: Detect and extract shadows from the at least three digital images; Estimate the direction of the selected shadow in the extracted shadows; and Individual rock fragments in the rock fragments are identified based on the calculated surface normal vector; and Individual rock fragments within the rock fragments are identified based on the estimated shadow direction.

5. The method according to claim 1, wherein generating the segmented image further comprises: A first segmented image is generated, in which each rock fragment is identified based on the surface normal vector calculated in the photometric stereo image; A second segmentation image is generated, in which individual rock fragments are identified based on the estimated direction of shadows extracted from the at least three digital images; The first segmented image and the second segmented image are combined to obtain the third segmented image; Apply edge detection or region growing techniques to the third segmentation image; and Estimating the features of the underground strata from the third segmented image.

6. The method of claim 1, wherein the acquisition, the combination, and the generation are performed automatically.

7. A system for generating segmented images of rock cuttings, the system comprising: A sample holder configured to receive rock cuttings; A digital camera is positioned and configured to record digital images of rock fragments on the sample holder; A light source configured to illuminate the sample holder; and A controller configured to (i) cause the digital camera to acquire at least three digital images of the rock fragments at corresponding non-coplanar illumination angles; (ii) combine the at least three digital images to generate a photometric stereo image of the rock fragments; and (iii) generate a segmented image that identifies individual rock fragments within the rock fragments.

8. The system of claim 7, wherein the controller is further configured to rotate the sample holder to at least three different angular orientations corresponding to the non-coplanar illumination angle when at least three digital images of the rock cuttings are acquired.

9. A method for generating a segmented image of rock debris particles, the method comprising: Obtain and prepare rock cuttings for imaging; The prepared rock cutting particles are placed in front of a digital camera; At least three digital images of the rock fragments were acquired at the corresponding non-coplanar illumination angles; The at least three digital images are combined to generate a photometric stereoscopic image of the rock fragments; A first segmented image is generated, in which each rock fragment is identified based on the surface normal vector calculated in the photometric stereo image; and A second segmentation image is generated, in which individual rock fragments are identified based on the estimated direction of shadows extracted from the at least three digital images; and The first segmented image and the second segmented image are combined to obtain the third segmented image.

10. The method according to claim 9, wherein: The placement includes placing the prepared drilling cuttings particles on a rotatable stage in front of the digital camera; and The acquisition includes rotating the rotatable stage to at least three different angular orientations and acquiring digital images of the rock fragments at each of the at least three different angular orientations.

Images