Rock crack recognition method based on depth image feature extraction

Abstract

The application discloses a rock crack recognition method based on depth image feature extraction and relates to the technical field of rock crack recognition, and comprises the following steps: constructing a continuous energy response tracing band, converting the gray level fluctuation, edge offset and texture direction of a rock crack image into time sequence energy flow lines to form an energy expression structure; based on the direction stable area formed in the continuous energy response tracing band, extracting the micro-displacement trajectory of the crack edge, performing difference stripping on the energy peak value area and the adjacent low response area to generate a crack energy difference atlas, and establishing the corresponding relationship between the energy flow lines and the difference atlas. Through the construction of the continuous energy response tracing band and the time sequence energy flow lines, the application realizes dynamic tracking and independent expression of crack energy, avoids energy aliasing, improves crack recognition precision, and realizes self-adaptive balance of the energy field through the crack direction migration sequence and the breathing type energy regulation, thereby enhancing the stability and reliability of the recognition result.

Description

Technical Field

[0001] This invention relates to the field of rock crack recognition technology, and more specifically to a rock crack recognition method based on depth image feature extraction. Background Technology

[0002] Rock crack identification based on depth image feature extraction refers to a technical approach that, based on imaging the surface or internal structure of rocks, utilizes the automatic representation and discrimination capabilities of deep image features to accurately identify and distinguish cracks present in rocks. This method no longer relies on manually set simple grayscale, edge, or texture indicators. Instead, it extracts deep features from multi-scale structural information, local morphological changes, and spatial differences between cracks and intact rock masses in rock images. This allows for the full representation of the crack's elongated shape, discontinuous direction, width variations, and bifurcation characteristics in the feature space, thereby effectively distinguishing real cracks from noise textures and surface defects. In this way, the location, extent, and evolution characteristics of cracks can be stably acquired even under complex backgrounds, uneven lighting, or variable crack morphologies, providing reliable technical support for rock mass stability analysis, engineering safety assessment, and geological disaster early warning.

[0003] The existing technology has the following shortcomings: In existing technologies, when extracting depth features from rock crack images, if cracks in the monitored area exhibit high-density distribution and are continuously expanding, the grayscale changes, edge responses, and texture gradients generated by multiple cracks during their evolution will create overlapping image energy distributions in space. This causes the crack energy trajectories, which should be distinguishable, to become coupled. As a result, the depth feature layer is prone to incorrectly fusing local features of multiple independent cracks when constructing structural representations, leading to the generation of superimposed structures lacking true physical correspondence. These superimposed structures are then misjudged by the model as the main crack morphology with a continuous direction. This misjudgment further results in the incorrect extraction of the crack's main propagation direction, causing directional deviations in subsequent disaster trend analysis based on crack orientation. This poses a potentially serious safety hazard in key application scenarios such as rock mass instability assessment and risk zoning determination.

[0004] The information disclosed in the background section is only intended to enhance the understanding of the background of this disclosure, and therefore may include information that does not constitute prior art known to those skilled in the art. Summary of the Invention

[0005] The purpose of this invention is to provide a rock crack identification method based on depth image feature extraction, so as to solve the problems in the background art mentioned above.

[0006] To achieve the above objectives, the present invention provides the following technical solution: a rock crack identification method based on depth image feature extraction, comprising the following steps: A continuous energy response recording band is constructed to convert the grayscale fluctuations, edge offsets and texture directions of rock crack images into time-series energy streamlines, forming an energy expression structure with time-trackable characteristics. Based on the directional stable region formed in the continuous energy response recording band, the micro-displacement trajectory of the crack edge is extracted, the energy peak region and the adjacent low response region are differentially separated to generate a crack energy difference map, and the correspondence between the energy streamline and the difference map is established. In the crack energy difference spectrum, a multi-directional interference isolation zone is set along the energy streamline direction. A time series reverse shift operation is performed on the energy overlap area. The energy streamline separation effect is generated by the time axis misalignment, and mutually independent energy channels are obtained. A crack orientation separation chain is established based on independent energy channels. The crack propagation direction in each energy channel is continuously tracked, and the location of abrupt change in direction is recorded as an unstable node, forming a crack orientation migration sequence. The energy field is controlled by breathing based on the crack direction migration sequence. By performing intermittent expansion and convergence operations on the energy streamlines, the residual coupling energy between cracks is periodically released, resulting in crack identification results with adaptive balance and dynamic stable separation characteristics in the time dimension.

[0007] Preferably, the steps for constructing the continuous energy response recording band are as follows: After acquiring the image of the rock crack to be identified, gray-level fluctuations are identified by measuring the changes in brightness levels in the image. The energy fluctuation differences of each gray-level unit in adjacent regions are determined by continuous measurement, and a gray-level response sequence with time extension is formed. After obtaining the grayscale response sequence, the offset path of the crack edge is extracted based on the boundary change trend of the grayscale change region, and the edge offset trajectory is continuously matched with the grayscale response sequence so that the crack edge energy change and grayscale energy fluctuation form a corresponding relationship in the time dimension. After obtaining the grayscale energy change and edge offset energy path, the texture direction of the rock crack image is continuously identified, and the texture energy continuation, grayscale response sequence and edge offset trajectory are uniformly calibrated to form a continuous energy flow field. After the three types of energy information are integrated in the time dimension, the fused energy flow field is continuously depicted to form a continuous energy response recording band covering the crack area, and an energy expression structure with time-trackable characteristics is established.

[0008] Preferably, the steps for generating a crack energy difference spectrum based on the directional stable region formed in the continuous energy response recording band are as follows: After obtaining the continuous energy response recording band, the changes in the amplitude and direction of energy response fluctuations are identified along the time extension direction of the energy streamline. When the energy streamline maintains a consistent direction in the spatial range, the directionally stable region is determined and used as the stable segment of crack energy change. Within the directional stability region, the micro-displacement trajectory of the crack edge is extracted along the energy streamline direction, and the edge energy change is continuously correlated with the gray-scale energy fluctuation, so that the crack edge energy response has continuous trajectory characteristics. After obtaining the micro-displacement trajectory of the crack edge, the energy peak region and the adjacent low response region in the directional stable region are differentially stripped to clearly separate the energy distribution boundary. After energy difference stripping, the stripped energy data is structured and integrated to generate a crack energy difference map, and a correspondence between energy streamlines and the difference map is established.

[0009] Preferably, in the process of generating the crack energy difference map, the energy gradient between the energy peak region and the adjacent low response region is sequentially stripped along the energy streamline direction. By extending the time dimension, the spatial energy difference is transformed into a temporal separation relationship, so that the energy of the crack body and the background energy form a clear boundary, and a continuous mapping relationship is maintained between the energy streamline and the difference map, so as to ensure the temporal traceability and spatial hierarchy of the crack energy distribution.

[0010] Preferably, the steps for setting multi-directional interference isolation zones along the energy streamline direction in the crack energy difference spectrum and obtaining mutually independent energy channels are as follows: Identify continuous regions of energy distribution along the energy flow lines, and set up multi-directional interference isolation zones at the intersection of energy flow lines to form a spatial isolation layer along the energy flow path; Based on the energy distribution boundary defined by the multi-directional interference isolation zone, the energy overlap region in the crack energy difference spectrum is identified and directionally divided within the isolation zone, so that the interference of the energy overlap region is restricted to an independent isolation channel; After the energy overlap region is defined, a time-series reverse shift operation is performed on it along the energy streamline direction. The energy streamline separation effect is generated by the time axis misalignment, thereby decoupling the energy coupling relationship. After the energy streamlines are misaligned due to the reverse shift of the time series, the misaligned and separated energy streamlines are integrated to form continuous and independent energy channels.

[0011] Preferably, the multi-directional interference isolation zone extends along the energy flow line direction and forms an energy buffer zone in the lateral direction. The energy buffer zone is used to limit the spatial interference of adjacent energy flow lines, so that the energy overlapping area maintains directional consistency when performing time series reverse shift operation. The continuity of energy flow lines after time axis misalignment is maintained by the boundary of the isolation zone, thereby ensuring the independent distribution and continuous transmission of energy channels.

[0012] Preferably, the steps for establishing a crack orientation separation chain and forming a crack direction migration sequence based on independent energy channels are as follows: After obtaining independent energy channels, the energy flow lines in each energy channel are identified along the time extension direction to determine the directional evolution trend of the energy flow lines, and the energy flow lines with the same direction are summarized as crack direction units. After identifying the crack direction units in each independent energy channel, the direction units are connected in an orderly manner, and a crack direction separation chain is established sequentially along the time axis so that the crack direction change is continuously recorded in the time dimension. In the crack propagation separation chain, nodes with a rate of change of direction higher than the average propagation rate are identified, and the locations of abrupt changes in direction are marked as unstable nodes, forming an unstable node cluster. After the location of the directional abrupt change is determined, the crack direction separation chain is time-series integrated to form a crack direction migration sequence, so that the crack propagation direction forms a continuous migration trajectory in the time dimension.

[0013] Preferably, the steps for controlling the energy field in a breathing manner based on the crack direction migration sequence are as follows: After obtaining the crack direction migration sequence, the overall distribution of the energy field is initially balanced according to the crack direction change trend, and a time period framework for energy flow is established so that the energy streamlines maintain an extension relationship consistent with the crack direction migration sequence. After the energy field achieves initial equilibrium, the energy streamlines are intermittently expanded according to the directional extension segment in the crack direction migration sequence, so that the streamlines of each independent energy channel expand outward and maintain directional consistency. After the intermittent expansion of the energy streamlines is completed, an intermittent convergence operation is performed based on the unstable node positions recorded in the crack direction migration sequence to guide the energy streamlines back towards the center and promote the release of residual coupled energy. After the energy streamlines undergo periodic operations of intermittent expansion and convergence, the energy field is dynamically integrated to obtain crack identification results that possess adaptive equilibrium and dynamic stable separation characteristics in the time dimension.

[0014] The technical effects and advantages provided by the present invention in the above technical solution are as follows: This invention introduces continuous energy response recording bands and time-series energy streamlines into the crack identification process, enabling a coherent representation of grayscale fluctuations, edge offsets, and texture orientation in rock crack images over time, and allowing dynamic tracking of energy distribution changes. The continuous mapping of energy along the time axis effectively avoids energy confusion caused by the superposition of spatial features, maintaining the independence of crack energy trajectories, thereby improving the ability to distinguish crack features under complex background conditions and ensuring the accuracy of crack morphology and orientation identification.

[0015] This invention establishes a crack direction migration sequence and implements breathing-like regulation of the energy field, enabling the residual coupling energy between cracks to be dynamically released during periodic expansion and convergence, achieving adaptive equilibrium of the energy field in the time dimension. Under continuous regulation, the energy channels maintain independent distribution, enhancing the directional stability and structural integrity of crack identification results. The identification process possesses dynamic separation characteristics, maintaining continuity and reliability even under high-density crack conditions. Attached Figure Description

[0016] To more clearly illustrate the technical solutions in the embodiments of this application or the prior art, the drawings used in the embodiments will be briefly introduced below. Obviously, the drawings described below are only some embodiments recorded in this invention. For those skilled in the art, other drawings can be obtained based on these drawings.

[0017] Figure 1 This is a flowchart of the rock crack identification method based on depth image feature extraction according to the present invention. Detailed Implementation

[0018] Exemplary embodiments will now be described more fully with reference to the accompanying drawings. However, these exemplary embodiments can be implemented in many forms and should not be construed as limited to the examples set forth herein; rather, they are provided so that the description of this disclosure will be more complete and fully convey the concept of the exemplary embodiments to those skilled in the art.

[0019] This invention provides, for example Figure 1 The rock crack identification method based on depth image feature extraction shown includes the following steps: A continuous energy response recording band is constructed to convert the grayscale fluctuations, edge offsets and texture directions of rock crack images into time-series energy streamlines, forming an energy expression structure with time-trackable characteristics. To obtain a temporally traceable energy representation structure from rock crack images, energy feature transformation is performed on the rock crack images to construct continuous energy response bands. The specific steps are as follows: After acquiring the image of the rock crack to be identified, grayscale fluctuations are identified based on the brightness level variations distributed throughout the image. By continuously measuring the grayscale change trend across the entire image range, the energy fluctuation difference of each grayscale unit in adjacent regions is determined. In this way, the brightness changes of each pixel in the image no longer exist merely as static differences, but are transformed into a grayscale response sequence with time-dependent characteristics. The construction of this grayscale response sequence provides an initial continuous input for subsequent energy conversion, allowing the grayscale changes in each local region to be described as an energy response curve flowing over time, thus providing a dynamic basis for the subsequent establishment of energy streamlines.

[0020] After obtaining the grayscale response sequence, the offset path of the crack edge is extracted based on the boundary change trend of the grayscale variation region. This step captures the subtle movement trajectory of the crack edge in the image structure by continuously tracking the spatial offset of the boundary line along the grayscale change direction. By continuously matching these edge offset trajectories with the grayscale response sequence obtained in the previous stage, a one-to-one correspondence is established between the energy change of the crack edge and the grayscale energy fluctuation in the time dimension. In this process, each crack edge is assigned a continuously changing energy path, giving it temporal coherence in the energy space, thus enabling the edge response to have dynamically traceable characteristics.

[0021] After obtaining the grayscale energy changes and edge offset energy paths, the texture orientation in the rock crack image is continuously identified. By spatially tracing the linear extension of the texture arrangement along the crack edge direction, the energy continuity trend of the texture orientation in the time dimension is determined. The energy continuity of the texture orientation is uniformly calibrated with the aforementioned grayscale response sequence and edge offset trajectory to ensure that the three feature dimensions form a consistent energy flow direction on the time axis. In this way, the energy changes of grayscale fluctuations, the energy paths of edge offsets, and the energy extension of texture orientation in the image are integrated into a continuously changing energy flow field. This energy flow field exhibits traceable flow characteristics in the time dimension, providing directional constraints for the stability of subsequent energy mapping.

[0022] After integrating the three types of energy information—grayscale fluctuations, edge offsets, and texture orientation—into a temporal dimension, the fused energy flow field is continuously depicted, forming a complete continuous energy response recording band. By connecting different energy components at relatively stable intervals along the time axis, the transmission relationship between energy streams remains continuous, thus forming a continuous energy flow trajectory throughout the entire image area. Each energy stream represents a continuously changing energy expression unit in the temporal dimension. The energy streams maintain a correspondence according to the directionality of grayscale energy fluctuations, the trend of edge offsets, and the extensibility of texture orientation, ultimately forming an energy expression structure covering the entire crack area. This energy expression structure is coherent and traceable in the time series, enabling a dynamic depiction of the entire process of energy changes in rock crack images, providing a stable energy basis for subsequent crack energy difference extraction, interference isolation, and orientation separation.

[0023] Based on the directional stable region formed in the continuous energy response recording band, the micro-displacement trajectory of the crack edge is extracted, the energy peak region and the adjacent low response region are differentially separated to generate a crack energy difference map, and the correspondence between the energy streamline and the difference map is established. To separate the main directional features of crack energy from continuous energy response recordings and establish a comparative relationship of energy distribution, feature extraction and energy difference construction are performed on the directionally stable regions formed in the energy response recordings. The specific steps are as follows: After obtaining the continuous energy response recording band, the spatial distribution range of the directionally stable region is determined for areas in the recording band where the energy streamlines exhibit directional consistency and temporal continuity. By identifying the fluctuation amplitude and directional shift of the energy response along the temporal extension direction of the energy streamlines, the region is marked as the directionally stable region when the continuous energy streamlines maintain a basically consistent energy direction within a certain spatial range and the rate of change is lower than a preset stability threshold. The formation of the directionally stable region means that the energy flow direction of the crack is in a relatively stable state within this local area, and the spatial trend of the crack edge shows a smooth extension in the time dimension. By determining the directionally stable region, stable segments of crack energy change can be clearly identified in the energy expression structure, providing a stable reference baseline for subsequent extraction of the micro-displacement trajectory of the crack edge, making the tracking of energy features continuous and directionally constrained.

[0024] After determining the directionally stable region, the micro-displacement trajectories of the crack edge are extracted based on the continuous distribution characteristics of energy streamlines within the stable region. By refining and tracing the high-energy response path representing the crack edge along the energy streamline direction, the slight changing trends of energy intensity in the time dimension are identified. The crack edge often appears as a slender region with dense and frequently changing energy in the continuous energy response recording band. Therefore, when extracting its displacement trajectory in the directionally stable region, it is necessary to maintain continuous tracing consistent with the energy streamline direction, so that each edge energy change is recorded as a displacement trajectory unit in the time series. Through the continuous extraction of these micro-displacement trajectories, the subtle dynamic changes of the crack edge in the energy flow field can be accurately described, allowing the local energy form of the crack to be extended and quantified in time, thus forming the spatiotemporal basic structure of crack energy changes. This step concretizes the implicit energy movement laws in the directionally stable region, giving the energy response of the crack edge continuous trajectory characteristics, and providing complete displacement information for energy difference analysis.

[0025] After obtaining the micro-displacement trajectory at the crack edge, a differential separation process is performed between the energy peak region and the adjacent low-response region within the directionally stable zone. By identifying regions with concentrated and continuously varying energy intensity along the energy trajectory at the crack edge, these are designated as energy peak regions; simultaneously, the weaker energy responses on either side of the peak regions are identified as low-response regions. Then, along the displacement trajectory direction in the time dimension of the energy streams, the energy gradient between the high and low energy regions is sequentially separated, clearly delineating the energy distribution boundary between the peak energy region and the adjacent low-response region. This differential separation process utilizes the temporal extension of continuous energy streams to transform spatial energy differences into temporal separation, thereby avoiding interference and overlap between energy peaks and background energy. The separated energy data exhibits clear energy boundary characteristics, clearly distinguishing the energy intensity of the crack body region from the surrounding background energy, and maintaining stable energy contrast in the time series, providing a clear differential basis for the subsequent generation of crack energy maps.

[0026] After differentiating between the energy peak region and the low response region, the separated energy data is structurally integrated to generate a crack energy difference map, and a correspondence is established between energy streamlines and the difference map. By preserving the mapping relationship between energy intensity distribution, direction information, and displacement trajectory in the time-series energy streamlines, the continuity of the energy streamlines is combined with the spatial partitioning of the difference map, so that each energy streamline corresponds to a specific energy difference unit in the map. The crack energy difference map, with the energy peak region as the core and the low response region as the background, forms a clear hierarchical structure of energy distribution, reflecting the energy difference state between the crack edge and the background region. By establishing the correspondence between energy streamlines and the difference map, a unified expression of temporal continuity and spatial hierarchy can be achieved in subsequent energy analysis and crack feature extraction. This correspondence not only ensures the traceability of crack energy distribution across different time periods but also provides a stable reference benchmark for energy coupling decoupling and direction separation, enabling subsequent processing to accurately identify crack evolution characteristics within the energy framework of the difference map.

[0027] In the crack energy difference spectrum, a multi-directional interference isolation zone is set along the energy streamline direction. A time series reverse shift operation is performed on the energy overlap area. The energy streamline separation effect is generated by the time axis misalignment, and mutually independent energy channels are obtained. To eliminate energy interference caused by the superposition of multiple crack energies in the crack energy difference spectrum and to ensure the spatial independence and temporal continuity between energy streamlines, the energy streamlines are separated. The specific steps are as follows: In crack energy difference maps, continuous regions of energy distribution are identified along the energy streamline direction, and multi-directional interference isolation zones are established at the intersections of these continuous regions. By determining the spatial proximity between different crack energy streamlines along their direction, an energy isolation zone is established at the center of the intersection region when multiple energy streamlines exhibit directional convergence or overlapping energy peaks within the same region. This energy isolation zone, constrained by the directional characteristics in the crack energy difference map, extends spatially along the energy streamline direction and forms an energy buffer zone of a certain width in the lateral direction, thus creating a directional dividing boundary in the energy flow field. By establishing multi-directional interference isolation zones in the crack energy difference map, a spatial isolation layer can be formed along the energy flow path, allowing different crack energy streamlines to maintain relatively independent propagation channels over subsequent time, preventing further spatial overlap of energy.

[0028] After establishing a multi-directional interference isolation zone, the energy overlap region in the crack energy difference spectrum is identified based on the energy distribution boundary defined by the isolation zone. The energy overlap region is typically located within the transition zone where two or more energy streamlines approach each other, and its energy characteristics are characterized by closely spaced energy peaks, abrupt directional changes, or concentrated grayscale gradient changes. By performing longitudinal analysis of the crack energy difference spectrum along the energy streamline direction, the temporal extent of the overlap region can be clearly defined. Then, based on the multi-directional interference isolation zone established in the previous stage, the energy data within the overlap region is directionally divided, confining each part of the energy distribution spatially within an independent isolation channel. In this way, the spatial interference of the energy overlap region is limited within the isolation zone, thus providing clear energy boundary conditions for subsequent time-series reverse shift operations.

[0029] After the energy overlap region is clearly defined, a time-series reverse shift operation is performed along the energy streamline direction. This operation uses time as a sequence reference, making subtle reverse adjustments to the response positions of the energy streamlines on the time axis to create a misalignment effect in the time dimension. Through the reverse shift of the time axis, the energy responses that originally overlapped at the same time point are dispersed to different time positions, thus achieving misalignment and separation of energy responses at the time level. During the reverse shift, the direction of the energy streamlines remains consistent with the main crack propagation direction to ensure the continuity of energy flow. The introduction of the time-series reverse shift dismantles the coupling relationship between energy streamlines in the time dimension, and the original interference relationship between the energy streamlines gradually weakens. Through this time axis misalignment operation, the energy streamlines maintain a consistent direction in space and form a small displacement difference in time, thereby achieving complete separation of energy responses in both the spatiotemporal dimensions, laying the foundation for constructing independent energy channels.

[0030] After the energy streamlines are misaligned through reverse time-series shifting, the misaligned energy streamlines are integrated to obtain independent energy channels. By reconstructing the time series of each misaligned energy segment along the energy streamline direction, each energy streamline forms a continuous energy channel on the time axis. The formation process of this energy channel maintains the overall directional characteristics of crack energy flow while avoiding interference with adjacent channels in terms of energy distribution. Multiple energy channels are spatially bounded by multi-directional interference isolation zones and temporally distinguished by the misaligned energy sequences, thus forming an independent yet continuous energy propagation structure in the entire crack energy difference map. In this way, the energy streamlines are transformed from their original coupled state into parallel and independent temporal energy channels, each maintaining stable energy transfer characteristics and fully describing the independent evolution process of the crack in the image energy space. The independence of energy channels not only eliminates the interference effects caused by the spatial superposition of crack energy but also constructs a traceable energy propagation path in the time dimension, providing stable energy support for subsequent crack orientation separation and direction migration analysis.

[0031] A crack orientation separation chain is established based on independent energy channels. The main propagation direction of the crack in each energy channel is continuously tracked. The location of the direction change is recorded as an unstable node, forming a crack direction migration sequence. To establish a crack direction tracking mechanism based on independent energy channels and to achieve continuous expression of crack direction over time, a crack direction migration sequence is formed through energy channel direction separation and dynamic tracking. The specific steps are as follows: After obtaining independent energy channels, the orientation of energy streamlines within each channel is identified. By continuously analyzing the spatial directional distribution of energy streamlines along the temporal extension direction of each independent energy channel, the directional evolution trend of each energy streamline on the time axis is determined. When multiple energy streamlines remain relatively parallel and oriented in the same energy channel, these streamlines are grouped into segments with the same orientation to form basic crack orientation units. This orientation unit represents the stable propagation direction of crack energy within a local region and reflects the energy transfer pattern of the crack within that energy channel over a short period. Through this orientation identification method, a preliminary crack orientation distribution framework can be established within independent energy channels, enabling the crack energy propagation relationship to have a traceable directional structure in the time series, providing a directional constraint basis for subsequent separation chain construction.

[0032] After identifying crack direction units in each independent energy channel, these units are sequentially connected to establish a crack direction separation chain. This chain connects adjacent direction units along the time axis, allowing for continuous recording of crack direction changes at different time stages. During the connection process, the continuity of the energy channel is maintained, preserving the spatial extension and energy transfer consistency of each direction unit. For direction units with similar directions in adjacent time periods, a sequential connection is made; for adjacent units with significant direction changes, the nodes of direction change are preserved during connection, ensuring the separation chain fully records the trajectory of crack direction changes over time. Through this continuous connection process, the crack direction separation chain not only forms a direction sequence throughout the entire energy channel but also structurally reflects the dynamic trend of crack energy flow, thus transforming static direction characteristics into a time-extendable direction chain expression, enabling the crack propagation direction to possess continuity and traceable evolution.

[0033] Based on the established crack orientation separation chain, nodes exhibiting directional changes within the chain are identified to pinpoint abrupt changes in crack orientation. By analyzing the directional shift of energy streamlines segment by segment along the crack orientation separation chain, nodes with directional change rates exceeding the average propagation rate are identified and marked as directional abrupt change locations. These locations often correspond to critical points where the crack undergoes turning, bifurcation, or propagation path deviation during energy transfer. In the crack energy channel, these locations represent regions where energy streamlines undergo directional redistribution, thus possessing significant structural change implications. After determining the directional abrupt change locations, the position coordinates, directional change trends, and energy response differences of these nodes are recorded chronologically, forming unstable node clusters. The extraction of unstable nodes provides dynamic identifiers of local changes to the crack orientation separation chain, in addition to its overall directional representation, allowing for the temporal expression of crack orientation abrupt changes. In this way, the directional change process of crack energy is refined into a series of continuous stable segments and intermittent directional abrupt change points, thus providing structured support for the dynamic tracking of crack evolution.

[0034] After the locations of directional abrupt changes are identified and marked as unstable nodes, the entire crack orientation separation chain is temporally integrated to form a crack orientation migration sequence. This sequence, with time as the main axis, arranges the directional segments and unstable nodes of the crack in different time periods according to their occurrence order, creating a continuous migration trajectory of the main crack propagation direction over time. This migration sequence reflects the directional extension law of crack energy at different time stages and reveals the dynamic changes of the crack in energy flow through the distribution of unstable nodes. Each crack orientation migration sequence corresponds to an independent energy channel; therefore, the entire crack energy system forms a directional set structure composed of multiple migration sequences. This structure extends continuously on the time axis and exhibits independent propagation in space, allowing the trajectory of crack orientation changes to be fully described in energy expression. Through the establishment of this orientation migration sequence, crack orientation is expanded from single-direction recording to multi-time-period dynamic tracking. The continuity, turning points, and instabilities of crack propagation are all quantitatively expressed in the time dimension, providing a traceable directional reference for subsequent energy field regulation stages.

[0035] The energy field is controlled by breathing based on the crack direction migration sequence. By performing intermittent expansion and convergence operations on the energy streamline, the residual coupling energy between cracks is periodically released, resulting in crack identification results with adaptive balance and dynamic stable separation characteristics in the time dimension. To dynamically balance and control the energy streamlines under the guidance of the crack migration sequence, and to maintain a stable separation of energy distribution between cracks, a breathing-style regulation of the energy field is implemented. The specific steps are as follows: After obtaining the crack direction migration sequence, the overall distribution of the energy field is initially balanced based on the crack direction change trends recorded in the migration sequence. By analyzing the temporal relationship between each directional segment and unstable node in the crack direction migration sequence, the energy concentration and sparse regions of the energy field at different time periods are determined. Based on the directional information of the energy concentration regions, the energy streamline distribution density within independent energy channels is temporally stratified, enabling the energy field to form a periodic energy fluctuation basis in the time dimension. The purpose of this step is to establish a temporal periodic framework for energy flow, giving subsequent expansion and convergence operations a controllable rhythm and spatial directional constraints, thereby laying a time reference for the periodic release of energy between cracks. Through this initial adjustment, the energy streamlines within the energy field maintain an extension relationship consistent with the crack direction migration sequence in the overall structure, ensuring that the energy flow direction and crack orientation changes remain coordinated.

[0036] After the energy field achieves initial equilibrium, intermittent expansion operations are performed on the energy streamlines according to the directional extension segments in the crack migration sequence. Intermittent expansion selectively increases the spatial spacing between energy streamlines in the time dimension, causing the streamlines of each independent energy channel to expand outward, thereby increasing the distance between energy flow paths. By performing expansion operations along the main direction of the crack migration sequence, the energy streamlines obtain adequate space for release, preventing re-coupling of energy between cracks due to excessive proximity during propagation. The continuity and directional consistency of the energy streamlines are maintained during expansion to ensure that energy continues to propagate along its original direction in the time dimension without disrupting the overall structure of the energy field. This step, through periodic expansion operations, allows the energy field to exhibit a regular expansion state in the time dimension, providing a spatial buffer channel for subsequent energy release.

[0037] After the intermittent expansion of the energy streamlines is completed, an intermittent convergence operation is performed on the energy streamlines based on the unstable node positions recorded in the crack direction migration sequence. The intermittent convergence operation corresponds to the expansion operation, and its purpose is to periodically guide the energy streamlines back towards the center in the energy field, causing energy to re-concentrate within a local area, thereby promoting the release of residual coupled energy between cracks. The convergence operation, centered on the unstable node, gradually guides the surrounding energy streamlines back in a temporal sequence, creating a short-term energy overlap zone in the local area, and gradually releasing residual energy over time. Through this periodic alternation of expansion and convergence, the energy field forms a breathing-like dynamic regulation mechanism, dispersing energy between cracks during expansion and releasing it during convergence, thus maintaining the dynamic balance of the overall energy field. This step enables the energy field to develop self-regulating characteristics in the time dimension, allowing the energy interaction between cracks to be gradually released in periodic fluctuations, preventing energy accumulation in local areas that could cause directional shifts or energy distortions.

[0038] After the energy streamlines undergo intermittent expansion and convergence, the entire energy field is dynamically integrated to obtain crack identification results with adaptive equilibrium and dynamic stability separation characteristics in the time dimension. By recording the periodic fluctuations of the energy streamlines, the energy flow states of the energy expansion and convergence stages are mapped to time nodes in the crack direction migration sequence, ensuring that the crack identification results maintain a characteristic distribution synchronized with the crack energy evolution in the time series. After the energy field completes the breathing-like regulation, the independence of energy channels between cracks is maintained for a long time, the temporal continuity of crack direction is strengthened, and energy propagation remains spatially separated and temporally balanced. The final crack identification results not only present a clear crack distribution structure in space but also reflect the dynamic equilibrium state of the crack propagation process in the time dimension. Through this breathing-like regulation mechanism, crack energy forms a dynamic cycle of periodic release and reconstruction during propagation, making the energy distribution tend to stabilize in time, and enabling the crack identification results to maintain clear, stable, and traceable characteristics even under complex rock mass conditions.

[0039] This invention introduces continuous energy response recording bands and time-series energy streamlines into the crack identification process, enabling a coherent representation of grayscale fluctuations, edge offsets, and texture orientation in rock crack images over time, and allowing dynamic tracking of energy distribution changes. The continuous mapping of energy along the time axis effectively avoids energy confusion caused by the superposition of spatial features, maintaining the independence of crack energy trajectories, thereby improving the ability to distinguish crack features under complex background conditions and ensuring the accuracy of crack morphology and orientation identification.

[0040] This invention establishes a crack direction migration sequence and implements breathing-like regulation of the energy field, enabling the residual coupling energy between cracks to be dynamically released during periodic expansion and convergence, achieving adaptive equilibrium of the energy field in the time dimension. Under continuous regulation, the energy channels maintain independent distribution, enhancing the directional stability and structural integrity of crack identification results. The identification process possesses dynamic separation characteristics, maintaining continuity and reliability even under high-density crack conditions.

[0041] The foregoing has only described certain exemplary embodiments of the present invention by way of illustration. Undoubtedly, those skilled in the art can modify the described embodiments in various ways without departing from the spirit and scope of the present invention. Therefore, the foregoing drawings and descriptions are illustrative in nature and should not be construed as limiting the scope of protection of the claims of the present invention.

Claims

1. A rock crack identification method based on depth image feature extraction, characterized in that, Includes the following steps: A continuous energy response recording band is constructed to convert the grayscale fluctuations, edge offsets, and texture orientation of rock crack images into time-series energy streamlines, forming an energy expression structure; The steps for constructing continuous energy response recordings are as follows: After acquiring the image of the rock crack to be identified, gray-level fluctuations are identified by measuring the changes in brightness levels in the image. The energy fluctuation differences of each gray-level unit in adjacent regions are determined by continuous measurement, and a gray-level response sequence with time extension is formed. After obtaining the grayscale response sequence, the offset path of the crack edge is extracted based on the boundary change trend of the grayscale change region, and the edge offset trajectory is continuously matched with the grayscale response sequence so that the crack edge energy change and grayscale energy fluctuation form a corresponding relationship in the time dimension. After obtaining the grayscale energy change and edge offset energy path, the texture direction of the rock crack image is continuously identified, and the texture energy continuation, grayscale response sequence and edge offset trajectory are uniformly calibrated to form a continuous energy flow field. After the three types of energy information are integrated in the time dimension, the fused energy flow field is continuously depicted to form a continuous energy response recording band covering the crack area, and an energy expression structure with time-trackable characteristics is established. Based on the directional stable region formed in the continuous energy response recording band, the displacement trajectory of the crack edge is extracted, the energy peak region and the adjacent low response region are differentially separated to generate a crack energy difference map, and the correspondence between the energy streamline and the difference map is established. The steps for generating a crack energy difference map based on the directional stable region formed in the continuous energy response recording band are as follows: After obtaining the continuous energy response recording band, the changes in the amplitude and direction of energy response fluctuations are identified along the time extension direction of the energy streamline. When the continuous energy streamline maintains a basically consistent energy direction within a certain spatial range and the rate of change is lower than the preset stability threshold, the region is marked as the direction stable region. The displacement trajectory of the crack edge is extracted along the energy streamline direction within the directional stability region, and the edge energy change is continuously correlated with the gray-scale energy fluctuation, so that the crack edge energy response has continuous trajectory characteristics. After obtaining the displacement trajectory of the crack edge, the energy peak region and the adjacent low response region in the directional stability region are differentially stripped to clearly separate the energy distribution boundary. After energy difference stripping, the stripped energy data is structured and integrated to generate crack energy difference map, and the correspondence between energy streamlines and difference map is established. In the crack energy difference spectrum, a multi-directional interference isolation zone is set along the energy streamline direction. A time series reverse shift operation is performed on the energy overlap area. The energy streamline separation effect is generated by the time axis misalignment, and mutually independent energy channels are obtained. A crack orientation separation chain is established based on independent energy channels. The crack propagation direction in each energy channel is continuously tracked, and the location of abrupt change in direction is recorded as an unstable node, forming a crack orientation migration sequence. The energy field is controlled by breathing based on the crack direction migration sequence. By performing intermittent expansion and convergence operations on the energy streamlines, the residual coupling energy between cracks is periodically released, and the crack identification result is obtained. The steps for breathing-style regulation of the energy field based on the crack direction migration sequence are as follows: After obtaining the crack direction migration sequence, the overall distribution of the energy field is initially balanced according to the crack direction change trend, and a time period framework for energy flow is established so that the energy streamlines maintain an extension relationship consistent with the crack direction migration sequence. After the energy field achieves initial equilibrium, the energy streamlines are intermittently expanded according to the directional extension segment in the crack direction migration sequence, so that the streamlines of each independent energy channel expand outward and maintain directional consistency. After the intermittent expansion of the energy streamlines is completed, an intermittent convergence operation is performed based on the unstable node positions recorded in the crack direction migration sequence to guide the energy streamlines back towards the center and promote the release of residual coupled energy. After the energy streamlines undergo periodic operations of intermittent expansion and convergence, the energy field is dynamically integrated to obtain crack identification results that possess adaptive equilibrium and dynamic stable separation characteristics in the time dimension.

2. The rock crack identification method based on depth image feature extraction according to claim 1, characterized in that, In the process of generating the crack energy difference map, the energy gradient between the energy peak region and the adjacent low response region is sequentially stripped along the energy streamline direction. By extending the time dimension, the spatial energy difference is transformed into a temporal separation relationship, so that the energy of the crack body and the background energy form a clear boundary, and a continuous mapping relationship is maintained between the energy streamline and the difference map.

3. The rock crack identification method based on depth image feature extraction according to claim 1, characterized in that, The steps for setting up multi-directional interference isolation zones along the energy streamline direction in the crack energy difference spectrum and obtaining mutually independent energy channels are as follows: Identify continuous regions of energy distribution along the energy flow lines, and set up multi-directional interference isolation zones at the intersection of energy flow lines to form a spatial isolation layer along the energy flow path; Based on the energy distribution boundary defined by the multi-directional interference isolation zone, the energy overlap region in the crack energy difference spectrum is identified and directionally divided within the isolation zone, so that the interference of the energy overlap region is restricted to an independent isolation channel; After the energy overlap region is defined, a time-series reverse shift operation is performed on it along the energy streamline direction. The energy streamline separation effect is generated by the time axis misalignment, thereby decoupling the energy coupling relationship. After the energy streamlines are misaligned due to the reverse shift of the time series, the misaligned and separated energy streamlines are integrated to form continuous and independent energy channels.

4. The rock crack identification method based on depth image feature extraction according to claim 3, characterized in that, The multi-directional interference isolation zone extends along the energy streamline direction and forms an energy buffer zone in the lateral direction. The energy buffer zone is used to limit the spatial interference of adjacent energy streamlines, so that the energy overlapping area maintains directional consistency when performing time series reverse shift operation. The continuity of energy streamlines after time axis misalignment is maintained by the boundary of the isolation zone.

5. The rock crack identification method based on depth image feature extraction according to claim 3, characterized in that, The steps for establishing a crack path separation chain and forming a crack direction migration sequence based on independent energy channels are as follows: After obtaining independent energy channels, the energy flow lines in each energy channel are identified along the time extension direction to determine the directional evolution trend of the energy flow lines, and the energy flow lines with the same direction are summarized as crack direction units. After identifying the crack direction units in each independent energy channel, the direction units are connected in an orderly manner, and a crack direction separation chain is established sequentially along the time axis so that the crack direction change is continuously recorded in the time dimension. In the crack propagation separation chain, nodes with a rate of change of direction higher than the average propagation rate are identified, and the locations of abrupt changes in direction are marked as unstable nodes, forming an unstable node cluster. After the location of the directional abrupt change is determined, the crack direction separation chain is time-integrated to form a crack direction migration sequence.

Images