Log storage, stacking and conveying method based on image recognition and related equipment

Abstract

The invention relates to a log storage stacking and conveying method and related equipment based on image recognition, and the method comprises the following steps: collecting a log image of a storage area through an industrial camera, extracting log contour data, and detecting the bending degree to obtain a bending parameter; and performing grade classification on the logs according to the bending parameters to generate a classification result, and marking different grades of logs in the log image to form a log marking image. And warehousing space planning is carried out based on the log marking images, stacking coordinate areas of logs of all grades are determined, and an optimized log conveying path is planned in combination with starting point and terminal point coordinates and obstacle detection. And finally, according to the priority sequence of the conveying paths, the mechanical arm is controlled to clamp the logs, the logs are conveyed to the corresponding stacking coordinate area along the planned path after being calibrated, and the technical problems that traditional log stacking and conveying mostly depend on artificial experience for recognition, and stacking is not standard due to judgment errors are solved.

Description

TECHNICAL FIELD

[0001] The present application relates to the field of image technology, and in particular to a log storage stacking and conveying method based on image recognition and related equipment. BACKGROUND

[0002] With the rapid development of wood processing industry, the automation and intelligence level of log storage and logistics has become a key factor affecting production efficiency and cost control. Traditional log stacking and conveying relies on manual experience for identification, classification and stacking, which not only has high labor intensity and low operation efficiency, but also easily leads to non-standard stacking due to judgment errors, affecting the continuity and stability of subsequent processing links. In recent years, image recognition and machine vision technology has been widely applied in industrial automation, providing a new technical path for realizing intelligent perception and decision-making in the log storage process. Through real-time image acquisition of logs by industrial cameras, combined with image processing algorithms to identify log features, it has become an important direction to improve the intelligent level of wood storage. SUMMARY

[0003] The present application aims to at least solve one of the technical problems existing in the prior art. To this end, the present application proposes a log storage stacking and conveying method based on image recognition and related equipment.

[0004] The technical solution adopted by the present application is: On the one hand, the present application provides a log storage stacking and conveying method based on image recognition, comprising the following steps: acquiring images of logs to be processed in the storage area by an industrial camera, obtaining log images; extracting log contour data from the log images, and detecting the bending degree of the logs based on the log contour data to obtain log bending parameters; classifying logs based on the log bending parameters to obtain log classification results, and marking logs of different grades in the log images based on the log classification results to generate log marked images; planning the space of the storage stacking area based on the log marked images to obtain stacking coordinate areas, and planning the conveying path of the logs based on the stacking coordinate areas to obtain log conveying paths; According to the log conveying path, the logs are clamped and moved to the stacking coordinate area by a mechanical arm.

[0005] Further, the acquiring images of logs to be processed in the storage area by an industrial camera, obtaining log images, comprises: dividing and parallel shooting the logs to be processed in the storage area by an industrial camera to obtain shooting images, and marking the shooting images according to the area where the logs are located to obtain partitioned and numbered images; Synchronously profile segmenting logs in each partition based on the partition number image to obtain a single tree image, and performing region integrity verification on the single tree image to obtain a log image.

[0006] Further, the log profile data in the log image is extracted, and a bending degree of the log is detected based on the log profile data to obtain a log bending parameter, including: Performing a gray scale enhancement processing on the single tree image in the log image to obtain an enhanced single tree image, and identifying a log edge point in the enhanced single tree image; Connecting the log edge point by profile fitting to generate a log initial profile, and eliminating redundant points in the log initial profile to obtain log profile data; Drawing a straight line based on two end points of the log profile data to obtain a log reference straight line, and calculating a distance parameter of each pixel point on the log profile data to the log reference straight line; Extracting a maximum bending offset in the distance parameter, and calculating a bending degree ratio based on a length of the log reference straight line and the maximum bending offset, and taking the bending degree ratio as a log bending parameter.

[0007] Further, based on the log classification result, different grades of logs in the log image are marked to generate a log marked image, including: Associating and matching the grade identifier in the log classification result with the single tree image in the log image to obtain a grade-image correspondence table, and labeling a grade number for the single tree image in the grade-image correspondence table to obtain a numbered single tree image; Determining a profile boundary of different grades of logs based on the numbered single tree image, and drawing a differentiating color frame for the profile boundary to obtain a color frame single tree image; Batch splicing the color frame single tree image according to the partition layout of the log image to obtain a partition marked image, and adding a partition name and a shooting time stamp to the partition marked image to obtain a log marked image.

[0008] Further, based on the log marked image, a space planning is performed on a storage stacking area to obtain a stacking coordinate area, including: Performing a grade color recognition on the color frame single tree image in the log marked image to obtain a grade color distribution, and counting a quantity of logs of each grade based on the grade color distribution; Calculating a storage space occupation amount of logs of each grade based on the quantity, and performing a capacity allocation calculation on the storage space occupation amount according to an area of a storage area to obtain an allocated storage area; The allocation storage area is subjected to stacking area segmentation by region division grid to obtain hierarchical stacking areas, and the hierarchical stacking areas are subjected to position arrangement according to preset grade priorities to obtain arranged stacking areas. Region coordinate points are established based on the arranged stacking areas, and the region coordinate points are subjected to boundary range calibration to obtain stacking coordinate regions.

[0009] Further, the log is subjected to conveying path planning based on the stacking coordinate regions to obtain a log conveying path, which comprises: The starting point coordinates of the single log in the log marking image are located, and the storage end point coordinates of the corresponding grade log are determined based on the stacking coordinate regions; A straight line connection path is drawn in the warehouse area two-dimensional map based on the starting point coordinates and the storage end point coordinates to obtain an initial conveying path, and the initial conveying path is subjected to warehouse area obstacle detection to obtain path obstacle points; The initial conveying path is subjected to obstacle adjustment based on the path obstacle points to obtain an optimized conveying path, and the optimized conveying path is subjected to priority sorting according to the moving distance to obtain the log conveying path.

[0010] Further, the log is gripped and moved to the stacking coordinate region by the mechanical arm according to the log conveying path, which comprises: The priority sorting information in the log conveying path is associated and bound with the grade identification in the corresponding log marking image to obtain a path-grade association table, and the path-grade association table is grouped according to the job batch to obtain a batch job list; The mechanical arm is called based on the batch job list, the position of the log corresponding to the first path in the batch job list is calibrated to obtain a calibration gripping point, and the mechanical arm is controlled to move to the calibration gripping point according to the log conveying path to obtain a to-be-gripped state; The mechanical arm in the to-be-gripped state is given a gripping instruction to grip the fixed object, and the mechanical arm is driven to move the gripped fixed object to the corresponding stacking coordinate region based on the log conveying path.

[0011] The application also provides a log warehouse stacking conveying device based on image recognition, which comprises: The acquisition module is used for image acquisition of logs to be processed in the warehouse area by an industrial camera to obtain log images; The extraction module is used for extracting log contour data in the log image, and detecting the bending degree of the log based on the log contour data to obtain log bending parameters; The classification module is configured to classify logs based on the log bending parameter, to obtain a log classification result, and to mark logs of different grades in the log image based on the log classification result, to generate a log marked image. The planning module is configured to plan a space of a storage and stacking area based on the log marked image, to obtain a stacking coordinate area, and to plan a conveying path of logs based on the stacking coordinate area, to obtain a log conveying path. The transferring module is configured to pick up and transfer logs to be processed to the stacking coordinate area by a mechanical arm according to the log conveying path.

[0012] The present application also provides a computer device comprising a memory and a processor, wherein the memory stores a computer program, and the processor executes the computer program to implement the steps of the method.

[0013] The present application also provides a computer readable storage medium storing a computer program, wherein the computer program is executed by a processor to implement the steps of the method.

[0014] The present application provides a log storage and stacking conveying method based on image recognition, comprising the following steps: acquiring images of logs to be processed in a storage area by an industrial camera, to obtain log images; extracting log contour data in the log images, and detecting the bending degree of logs based on the log contour data, to obtain log bending parameters; classifying logs based on the log bending parameters, to obtain a log classification result, and marking logs of different grades in the log images based on the log classification result, to generate a log marked image; planning a space of a storage and stacking area based on the log marked image, to obtain a stacking coordinate area, and planning a conveying path of logs based on the stacking coordinate area, to obtain a log conveying path; picking up and transferring logs by a mechanical arm to the stacking coordinate area according to the log conveying path, which solves the technical problem that traditional log stacking and conveying are mainly dependent on manual experience for identification, and are prone to cause non-standard stacking due to judgment errors, and uses the generated log marked image to combine the size and shape characteristics of logs of different grades to intelligently allocate a stacking area, to realize dynamic partitioning and efficient utilization of a storage area, to effectively improve the storage capacity per unit area and to guarantee the stability of the stacking structure. BRIEF DESCRIPTION OF DRAWINGS

[0015] The above and / or additional aspects and advantages of the present application will become apparent and be readily understood from the following description, taken in conjunction with the following drawings, in which: Figure 1 A step flowchart of the log storage and stacking conveying method based on image recognition in the embodiments of the present application; Figure 2is a structural block diagram of a log storage stacking and conveying device based on image recognition in an embodiment of the present application. Figure 3 is a structural schematic block diagram of a computer device in an embodiment of the present application.

[0016] The purposes, functional features and advantages of the present application will be further described with reference to the embodiments and the accompanying drawings. DETAILED DESCRIPTION

[0017] Embodiments of the present application are described in detail below, examples of which are shown in the accompanying drawings, wherein the same or similar notations represent the same or similar elements or elements having the same or similar functions throughout. The embodiments described below by reference to the accompanying drawings are exemplary and are only used to explain the present application, and cannot be understood as a limitation of the present application.

[0018] In the description of the present application, it should be understood that the orientation description, such as the orientation or position relationship indicated by up, down, front, back, left, right, etc. is based on the orientation or position relationship shown in the drawings, and is only for the convenience of describing the present application and simplifying the description, and does not indicate or imply that the device or element indicated must have a particular orientation, be constructed and operated in a particular orientation, and therefore cannot be understood as a limitation of the present application.

[0019] In the description of the present application, the meaning of several is one or more, and the meaning of multiple is more than two, greater than, less than, more than, etc. are understood as not including the number, and above, below, etc. are understood as including the number. If it is described as first, second, it is only used for the purpose of distinguishing technical features, and cannot be understood as indicating or implying relative importance or implicitly indicating the number of indicated technical features or the sequence of indicated technical features.

[0020] In the description of the present application, unless otherwise explicitly limited, the words such as setting, installing, connecting, etc. should be broadly understood, and the person skilled in the art can reasonably determine the specific meaning of the above words in the present application in combination with the specific content of the technical solution.

[0021] The embodiments of the present application are further described below with reference to the accompanying drawings.

[0022] Referring to Figure 1 The embodiment of the present application provides a log storage stacking and conveying method based on image recognition, comprising the following steps: Step S1, image acquisition of the logs to be processed in the storage area is performed by an industrial camera to obtain log images.

[0023] Specifically, the image of the raw wood to be processed in the storage area is collected by an industrial camera, and the raw wood image is obtained. This process first requires installing an industrial camera at a suitable position in the storage area to ensure that its field of view can cover the area of the raw wood to be processed. Then, according to the actual environmental lighting conditions, adjust the camera parameters such as exposure time, gain, etc. to ensure that the clarity and contrast of the collected image are suitable for subsequent analysis. Then, start the image acquisition program to capture the image data of the raw wood in real time. During this process, it may be necessary to consider continuous shooting or trigger shooting methods, and determine the best acquisition opportunity according to the speed and position of the raw wood entering the field of view. For example, in a typical wood storage scene, when a batch of raw wood is transported to the designated area for classification processing, the industrial camera will automatically take pictures at preset intervals or be manually triggered by the staff to obtain raw wood images containing as much detailed information as possible. These images will serve as the basis for subsequent steps to extract raw wood contour data and further process based on these data. In this way, by precisely controlling the image acquisition process, high-quality data support is provided for subsequent automated processing, ensuring the accuracy and efficiency of the entire process. At the same time, in order to adapt to the application requirements in different environments, it may be necessary to continuously adjust the camera settings and position to ensure that satisfactory image results can be obtained even in the case of changing light or irregularly placed raw wood.

[0024] Step S2, extracting the raw wood contour data in the raw wood image, and detecting the bending degree of the raw wood based on the raw wood contour data to obtain the raw wood bending parameter.

[0025] Specifically, the log profile data in the log image is extracted, and the log bending degree is detected based on the log profile data to obtain the log bending parameter. This process first performs grayscale processing on the obtained log image to reduce the calculation complexity and enhance the accuracy of subsequent edge recognition. Then, Gaussian filtering is used to denoise the image to eliminate interference caused by uneven lighting or camera noise. Then, the edge detection algorithm such as Canny or Sobel is used to identify the boundary information of each log in the image. Then, the contour extraction algorithm (such as the findContours method in OpenCV) is used to combine the continuous closed boundary points into independent profile data, and each profile corresponds to the outer shape of a log. Subsequently, for each extracted log profile data, the axis curve is fitted, usually using the least square method to fit the center line or the multi-segment straight line approximation method, and then calculating the maximum offset distance between the axis and its ideal straight line state. This distance is the core indicator to measure the bending degree. After normalization with the total length of the log, the specific log bending parameter is formed. For example, in an actual scenario of a wood storage area, when a log has a significant arc due to natural growth, its profile data after processing can accurately reflect that the central part of the axis deviates from the straight line reference by 8 cm. The system calculates the bending parameter as 8% based on this, and uses it as the classification basis to enter the next stage of processing, thereby ensuring that the bending quantification result is consistent with the actual shape and providing reliable data support for subsequent grade classification.

[0026] Step S3, classifying the logs based on the log bending parameter to obtain a log classification result, and marking logs of different grades in the log image based on the log classification result to generate a log marked image.

[0027] Specifically, the logs are classified based on the log bending parameters to obtain log classification results, and different grades of logs are marked in the log image based on the log classification results to generate a log marked image. This process first compares the log bending parameters obtained in the previous step with the preset grade division standard, which is set according to industry standards or internal requirements of the enterprise. For example, logs with a bending parameter less than or equal to 3% are classified as first-class logs, logs with a bending parameter greater than 3% and less than or equal to 8% are classified as second-class logs, and logs with a bending parameter greater than 8% are classified as third-class or out-of-specification products. The bending parameters of each log are determined by conditional judgment statements one by one, and the corresponding log classification results of each log are obtained. Subsequently, the contour area of each log is located in the original log image, and different colors or label marks are assigned according to the classification results, such as using a green rectangular frame to mark first-class logs, a yellow polygonal frame to mark second-class logs, and a red text label to mark third-class logs. The grade number or bending parameter value is superimposed at the corresponding position. All labeling operations are accurately superimposed based on the spatial coordinate information of the log contour data to ensure that the marked position corresponds to the actual log position one by one. For example, in the actual running scene of the warehouse area, when the system identifies that the bending parameter of a log is 5.2%, it is determined that the log belongs to the second-class log. A yellow frame is drawn around the contour of the log in the log image, and the word "second-class" is displayed above it. Finally, all the marked logs together form a complete log marked image for subsequent space planning.

[0028] Step S4, based on the log marked image, the space of the warehouse stacking area is planned to obtain the stacking coordinate area, and based on the stacking coordinate area, the conveying path of the logs is planned to obtain the log conveying path.

[0029] Specifically, based on the log marking image, the storage stacking area is spatially planned to obtain a stacking coordinate area, and based on the stacking coordinate area, a conveying path for the logs is planned to obtain a log conveying path. This process first needs to analyze the position information and grade classification results of each log in the log marking image, and through algorithm recognition and calculation, the ideal storage position of each grade of log in the storage area is calculated, for example, first-grade logs are placed in the area close to the entrance for quick access, and third-grade logs are arranged relatively far from the entrance. Then, according to the position information, the stacking coordinate area of each grade of log is determined, and the space filling algorithm is used to optimize the space utilization of these areas to ensure that each category of log has a clear storage limit. Next, after the stacking coordinate area is determined, combined with the warehouse layout map and the current position of the log, the path search algorithm such as A* algorithm is used to calculate the best conveying path for each log from the current position to the designated stacking coordinate area, considering factors including but not limited to shortest distance, collision avoidance, and priority sorting. For example, in an actual application, when the system processes a log marked as second grade, it will first determine that the stacking coordinate of this grade of log is located in the middle left area of the storage area, and then according to the specific coordinates of this log in the unloading area, a path that bypasses the working stacker and has the shortest total length is planned, ensuring that the log can be efficiently and accurately transported to the predetermined storage position. The entire process seamlessly connects to ensure the smoothness and efficiency of the storage operation.

[0030] Step S5, according to the log conveying path, the log is clamped and moved to the stacking coordinate area by the mechanical arm.

[0031] Specifically, according to the log conveying path, a robotic arm grips and moves the logs to the palletizing coordinate area. This process first loads the generated log conveying path data, which is planned based on the palletizing coordinate area in the previous steps and includes the complete movement trajectory from the current position of the log to the target palletizing position. The robotic arm control system reads the coordinates of key nodes in the path and performs real-time position calibration in conjunction with the positioning sensors deployed on-site to ensure motion accuracy. When the system confirms that the position of a log to be processed is aligned with the starting point of the conveying path, it controls the robotic arm to move along the predetermined trajectory to the pre-grabbing position above the log. Then, it adjusts the posture of the gripping device to match the spatial orientation of the log, determines the optimal gripping point based on the log contour data, and avoids damage or slippage. Next, the robotic arm performs a descent action, initiates the gripper closure to complete the gripping operation, and detects the gripping force feedback signal to confirm that the clamping state is stable. After gripping, the robotic arm moves step by step according to the obstacle avoidance route and speed parameters set in the log conveying path. During the process, the operating status is continuously monitored by an encoder and vision assistance system to prevent deviation from the trajectory or collision. For example, in actual operation in the storage area, when a secondary log is planned to be moved to the palletizing coordinate area slightly to the left of the center, the robotic arm will pick up the log from the unloading area according to the calculated conveying path, bypass other equipment in the adjacent working channel, move smoothly to the target stacking point, and finally accurately lower and release the gripper, placing the log in the designated position in the palletizing coordinate area, completing the entire transfer process.

[0032] In a specific scenario, the process of acquiring images of logs to be processed in the warehouse area using an industrial camera to obtain log images includes: Industrial cameras are used to take parallel photos of logs to be processed in the storage area in sections, and the photos are then numbered and marked according to the area where the logs are located, resulting in section-numbered images. Based on the partition number image, the logs in each partition are synchronously outlined to obtain individual log images, and the regional integrity of the individual log images is verified to obtain log images.

[0033] Specifically, the industrial cameras are used to take pictures of the logs in the storage area, and the pictures are numbered according to the regions where the logs are located. This process first divides the storage area into several physical regions with clear boundaries. The division of these regions is based on the layout of the storage, the density of the logs, and the field of view of the industrial cameras. This ensures that the distribution of logs in each region is neither too dense to cause image overlap nor too sparse to waste resources. Multiple industrial cameras or the same camera at different angles and positions are used to take pictures simultaneously or at different times to cover each region. This forms a set of pictures that contain the overall distribution of logs in different regions. Then, each picture is numbered according to its corresponding physical region. For example, the first region in the northeast corner of the storage area is labeled "01", and the second region in the southwest corner is labeled "02". The number is embedded in the image data using digital watermarking or file naming, generating a partitioned numbered image with location identification. This ensures that subsequent processing can accurately trace the spatial origin of each image. Next, the logs in each partition are profiled based on the partitioned numbered image. This step uses image processing algorithms to perform edge detection and connected component analysis on each partitioned numbered image independently. By setting appropriate thresholds and morphological operations, the algorithm separates logs that are in contact or partially obscured. It extracts the independent contour boundary of each log and crops the corresponding single log image segment from the original image, resulting in a single log image. In this process, parallel computing architecture is used to perform profile segmentation on multiple partitions simultaneously, improving overall processing speed and avoiding time delays caused by serial processing. Then, the single log image is checked for region integrity. This checks whether the log contour in each single log image is complete and whether there are any edge truncations or missing heads or tails. If a log's contour is incomplete in the current partition image, a retake mechanism is triggered or adjacent region images are called to perform image stitching to compensate for the missing parts. This ensures that the final single log image accurately reflects the actual shape of the log. Only single log images that pass the integrity check are retained and integrated into the final log image set, which serves as the basis for subsequent curvature detection and classification. For example, in an actual wood storage scenario, when a batch of logs is placed in three different regions of the storage area, three industrial cameras take pictures simultaneously to form three sets of pictures, which are labeled as "01", "02", and "03" partitioned numbered images. The system then performs profile segmentation on the images of these three regions in parallel. Five single log images are segmented from the "02" image, and their contour integrity is verified one by one. One log has a missing tail due to its proximity to the region boundary, so the system calls the adjacent image in the "01" region to complete the edge. After confirming the accuracy, it is included in the log image set, ensuring the accuracy and integrity of the image data relied upon by subsequent processing steps.

[0034] In a specific scenario, the log profile data in the log image is extracted, and the log is detected based on the log profile data to obtain a log bending parameter, including: The single tree image in the log image is subjected to gray scale enhancement processing to obtain an enhanced single tree image, and log edge points in the enhanced single tree image are identified; The log edge points are connected by profile fitting to generate a log initial profile, and redundant points in the log initial profile are removed to obtain log profile data; A straight line is drawn based on the two end points of the log profile data to obtain a log reference straight line, and a distance parameter of each pixel point on the log profile data to the log reference straight line is calculated; The maximum bending offset in the distance parameter is extracted, and a bending ratio is calculated based on the length of the log reference straight line and the maximum bending offset, and the bending ratio is taken as a log bending parameter.

[0035] Specifically, the single wood image in the log image is subjected to gray scale enhancement processing to obtain an enhanced single wood image, and the log edge points in the enhanced single wood image are identified. This process first performs image preprocessing operations on each single wood image that passes the integrity verification. Since the original image may be affected by uneven lighting, shadow interference or camera imaging quality, resulting in insufficient contrast of the log edge, it is necessary to use gray scale enhancement technology to improve the image quality. Specifically, histogram equalization or adaptive gamma correction method is used to expand the dynamic range of the image by nonlinear transformation of the pixel gray scale value, so that the boundary between the log and the background is clearer, thereby obtaining the enhanced single wood image. Subsequently, an edge detection algorithm such as Canny operator or Laplacian-of-Gaussian operator is applied to the enhanced image to identify the continuous edge points of the outer edge of the log by setting appropriate high and low thresholds. These edge points constitute the preliminary boundary set of the log in the image. Then, the log edge points are connected by contour fitting to generate the initial contour of the log. This step sorts the edge points and uses a curve fitting algorithm (such as least squares fitting or B-spline interpolation) to connect the discrete edge points into a closed or approximately continuous curve, forming the initial contour shape of the log. On this basis, the initial contour of the log is further optimized to remove redundant points. These redundant points are usually caused by image noise, knot reflection or adjacent log shielding, and are removed by setting a curvature change threshold or using the Douglas-Peucker algorithm to simplify the contour while retaining the main geometric features. Finally, the simplified and accurate log contour data is obtained. A straight line is drawn based on the two endpoints of the log contour data to obtain the log reference straight line. Specifically, the two most distant pixel points in the log contour data are extracted as the head and tail endpoints of the log, usually by calculating the two points with the maximum Euclidean distance in the contour point set, and then connecting these two endpoints to form a straight line that passes through the ideal axis of the log, which is the log reference straight line. This straight line represents the theoretical extension direction of the log in the straight state. Subsequently, the distance parameters of each pixel point on the log contour data to the log reference straight line are calculated. The perpendicular distance formula from a point to a straight line is used to calculate each point on the contour one by one to obtain the degree of deviation from the reference straight line. These distance parameters reflect the deformation distribution of the log along the length direction. The maximum bending offset in the distance parameters is extracted, which is the maximum value among all distance values. This value corresponds to the deviation degree of the most severe bending position of the log. Finally, the bending degree ratio is calculated based on the length of the log reference straight line and the maximum bending offset. Specifically, the maximum bending offset is divided by the total length of the log reference straight line and multiplied by 100% to obtain the bending degree value expressed in percentage, and the bending degree ratio is used as the log bending parameter for subsequent grade classification.For example, in an actual application of a timber storage area, the single wood image of a certain second-class raw wood is clearly displayed after gray scale enhancement, the system identifies the edge points and fits the initial contour, removes the redundant points caused by the reflection of tree bark, and obtains the accurate raw wood contour data. The length of the reference straight line is measured to be 4000 mm, and the maximum bending offset is 160 mm. The bending ratio is calculated to be 4.0%, which is used as the raw wood bending parameter and transmitted to the classification module to determine whether it belongs to the second-class raw wood category, thereby ensuring the accuracy of subsequent stacking and conveying decisions.

[0036] In a specific scenario, the raw wood classification result is used to mark different grades of raw wood in the raw wood image to generate a raw wood marked image, including: The grade identifier in the raw wood classification result is associated and matched with the single wood image in the raw wood image to obtain a grade-image correspondence table, and the single wood image in the grade-image correspondence table is labeled with a grade number to obtain a numbered single wood image; The contour boundary of different grades of raw wood is determined based on the numbered single wood image, and the contour boundary is drawn with a differentiated color box to obtain a color box single wood image; The color box single wood image is batch spliced according to the partition layout of the raw wood image to obtain a partition marked image, and a partition name and a shooting time stamp are added to the partition marked image to obtain a raw wood marked image.

[0037] Specifically, the grade identification in the log classification result is associated and matched with the single log image in the log image to obtain a grade-image correspondence table, and the single log image in the grade-image correspondence table is labeled with a grade number to obtain a numbered single log image. This process first establishes a data association between the log classification result obtained in the previous step, i.e., the grade information corresponding to each log (such as first grade, second grade, or third grade), and the single log image corresponding to it in the log image. This is achieved by using a unified number or unique identifier, for example, each single log image is assigned a number "02-03" (indicating the 3rd log in the 02 area) related to its region and position when it is generated, and the classification result also contains the same number of grade determination records. The system accurately corresponds the two through database query or key value matching method to form a grade-image correspondence table containing "image number-grade identification" entries, ensuring that the classification result of each log can be accurately traced back to its original image segment. Subsequently, based on the correspondence table, the numbered single log image is executed for each corresponding single log image, i.e., the grade information is superimposed on the image in text form at a prominent position such as the upper or side blank area of the log outline, using a fixed font, font size, and color (such as white bold font) to write "first grade", "second grade", etc. to generate a numbered single log image with visible numbers. This operation is completed through an image superposition function to ensure that the text does not cover the main body area of the log while maintaining clear readability. Next, based on the numbered single log image, the outline boundaries of logs of different grades are determined. This step uses the previously extracted and saved log outline data to obtain the closed boundary point set of each log in the image, then selects a preset differentiated color frame style according to the corresponding grade, for example, first grade logs use a green rectangular frame, second grade logs use a yellow polygon frame, and third grade logs use a red dashed line frame. The image drawing interface is called to accurately superimpose the boundary frame of the corresponding color on the periphery of the numbered single log image to form a color-framed single log image, making different grades visually distinct. Subsequently, the color-framed single log image is batch spliced according to the zoning layout of the log image, i.e., multiple color-framed single log images from the same storage zone are recombined into a complete regional view according to their original shooting position and spatial relative relationship. The splicing process preserves the physical spacing and arrangement order between images to avoid overlapping or misplacement, generating a zoning marker image that reflects the marking status of all logs in the region. Finally, the zoning marker image is added with a zone name and a shooting timestamp, for example, "02 Zone Log Marker Map" is written in the upper right corner of the image and "20250909_142315" is marked in the lower left corner to indicate the shooting time as September 9, 2025, 14:23:15. This information is presented in a semi-transparent background frame to prevent obscuring key image content. The final output is a complete, traceable, and highly visual log marker image.For example, in an actual timber storage scenario, after the system processes the classification results of the five logs in area 02, the single log images of "02-01" to "02-05" are labeled as "second class", "first class", "third class", "first class", and "second class" respectively, and the contours are marked with yellow, green, red, etc. Color frame, then horizontally splice the five color frame single log images into a whole view of area 02 according to the actual stacking order, add the name of "area 02" and the current timestamp, generate the final log marking image, and call it by the subsequent space planning module. Ensure that the stacking decision is based on accurate and intuitive visual data.

[0038] In a specific scenario, the space planning of the storage stacking area based on the log marking image to obtain the stacking coordinate area includes: The color frame single log image in the log marking image is subjected to grade color recognition to obtain a grade color distribution, and the number of logs of each grade is counted based on the grade color distribution. The storage space occupancy of logs of each grade is calculated based on the number, and the capacity allocation calculation of the storage space occupancy is performed according to the area of the storage area to obtain an allocated storage area. The allocated storage area is divided into stacking areas by region division grid to obtain hierarchical stacking areas, and the hierarchical stacking areas are arranged according to a preset grade priority to obtain arranged stacking areas. The region coordinate points are pre-established based on the arranged stacking areas, and the boundary range of the region coordinate points is calibrated to obtain the stacking coordinate area.

[0039] Specifically, the color frame single wood image in the original wood marking image is subjected to grade color recognition to obtain a grade color distribution, and the number of original woods of each grade is counted based on the grade color distribution. This process first extracts all single wood image regions with color frames from the generated original wood marking image, and uses image color segmentation technology, such as a threshold segmentation method based on the HSV color space, to identify preset differentiated color frames such as green, yellow and red. Green corresponds to first-grade original wood, yellow corresponds to second-grade original wood, and red corresponds to third-grade original wood. The system detects and classifies the position and number of each color frame by traversing the image pixel region to form grade color distribution data, such as identifying 3 green frames, 4 yellow frames and 2 red frames in the image, thereby counting 3 first-grade original woods, 4 second-grade original woods and 2 third-grade original woods. Then, the storage space occupancy of original woods of each grade is calculated based on the number. The calculation is based on the average length, diameter and ground projection area of each original wood according to the stacking method (such as vertical and horizontal interlacing or parallel arrangement) to estimate, for example, set the average occupancy of first-grade original wood as 1.8 square meters, second-grade as 1.7 square meters, and third-grade as 1.6 square meters. Then, the total occupancy of first-grade is 3 x 1.8 = 5.4 square meters, the total occupancy of second-grade is 4 x 1.7 = 6.8 square meters, and the total occupancy of third-grade is 2 x 1.6 = 3.2 square meters. The total storage space required is 15.4 square meters. Subsequently, the storage space occupancy is allocated according to the area of the storage area to obtain the allocated storage area. Specifically, combining the total available area of the storage area (such as 50 square meters) and the stacking priority of original woods of each grade (such as first-grade priority near the entrance), the total space is divided into several sub-areas by using proportional allocation or weighted allocation algorithm to ensure that the allocated storage area of original woods of each grade meets the actual demand and does not exceed the physical boundary. Then, the allocated storage area is divided into stacking areas by dividing the area into a grid, i.e. dividing the ground of the storage area into a plurality of square or rectangular grid units (such as each grid being 1m x 1m), determining the required number of grids according to the required allocated storage area of each grade, and combining continuous or adjacent grids into independent storage blocks to form first-grade stacking areas, second-grade stacking areas and third-grade stacking areas, i.e. to obtain hierarchical stacking areas. On this basis, the hierarchical stacking areas are arranged according to the preset grade priority, for example, first-grade original wood needs to be arranged in the front area near the conveyor belt exit due to high processing priority, second-grade original wood is arranged next, and third-grade original wood is arranged in the rear or corner area. The system adjusts the relative positions of the hierarchical stacking areas according to the coordinate system in the storage layout to avoid interference, and finally forms an orderly space layout, i.e. to obtain the arranged stacking area.Finally, the region coordinate point is established based on the arranged stacking region. This step sets the center point or corner point coordinate for each arranged stacking region in the global coordinate system of the storage area, for example, the first-level region is set as (X=10.0, Y=5.0), and the second-level region is set as (X=15.0, Y=6.0). The boundary range of the region coordinate point is calibrated by setting a safety distance and a stacking boundary, for example, a rectangular boundary is formed by extending outward 2.5 meters from the center point as the reference, to clearly define the limit range of the region allowed for stacking, to prevent out-of-bound stacking from affecting the passage of the channel or other equipment operations. Finally, the stacking coordinate region data containing accurate geographic coordinates is output. For example, in an actual wood storage scene, when the system processes the log marking image of the 02 area, it identifies that there are 2 first-level logs and 3 second-level logs, and calculates that 3.6 square meters and 5.1 square meters of storage space need to be allocated respectively. Then, 3x2 grids are divided in the front part of the storage area as the first-level arranged stacking region, and 4x2 grids are divided in the middle part as the second-level region, and the coordinate points with (8.0, 4.0) and (12.0, 5.0) as the centers are established. The boundary range is calibrated to ±1.2 meters, and finally the stacking coordinate region that can be used for path planning is generated, to ensure that the subsequent robotic arm can accurately move the logs to the specified position.

[0040] In a specific scenario, the wood conveying path is planned based on the stacking coordinate region, and the log conveying path is obtained, including: The starting point coordinate of a single log in the log marking image is located, and the storage end point coordinate of the corresponding level log is determined based on the stacking coordinate region; A straight line connection path is drawn in the two-dimensional map of the storage area based on the starting point coordinate and the storage end point coordinate, to obtain an initial conveying path, and the initial conveying path is detected for obstacles in the storage area to obtain path obstacle points; The initial conveying path is adjusted based on the path obstacle points to obtain an optimized conveying path, and the optimized conveying path is prioritized according to the moving distance to obtain the log conveying path.

[0041] Specifically, the starting point coordinates of each log in the log marking image are located, and the storage end point coordinates of the corresponding grade log are determined based on the stacking coordinate area. First, the spatial position information of each color frame single log image is extracted from the log marking image. The mapping relationship between the image coordinates and the actual physical coordinates of the storage area is used to convert the contour center point of the single log in the image into the actual starting point coordinates in the two-dimensional coordinate system of the storage area. For example, the camera projection matrix is calibrated to convert the image pixel coordinates (such as x=320, y=240) into ground coordinates (X=5.2, Y=3.8). The coordinates are the accurate position of the log. Then, according to the grade identification of the log in the log classification result, the stacking coordinate area matching the grade is found, for example, the coordinate range of the first-class log corresponding to the first-class stacking area. The system determines a specific storage end point coordinate from the coordinate range. The end point coordinate is usually selected as the center point of the idle grid closest to the entrance in the region that has not been occupied, to ensure the order of stacking and facilitate subsequent retrieval. Then, a straight line connection path is drawn in the two-dimensional map of the storage area based on the starting point coordinates and the storage end point coordinates, to obtain the initial conveying path. The two-dimensional map is a pre-constructed storage area plan, which contains all the fixed structures, channels, device positions and other information. The system draws a straight line segment on the map with the starting point and the end point as the endpoints, as the shortest theoretical moving path of the mechanical arm or conveying device. Then, the storage area obstacle detection is performed on the initial conveying path to obtain the path obstacle points. This step identifies whether the path crosses the stacker parking area, cable trench, other log stacking area or fixed column by superimposing and analyzing the initial conveying path and the obstacle layer in the two-dimensional map of the storage area. If there is an intersection, the coordinates of these conflict points are recorded as path obstacle points. For example, if it is detected that the initial path crosses the outline of a stationary forklift, the multiple coordinate points in the intersection area are marked as obstacle points. Based on the path obstacle points, the initial conveying path is adjusted to avoid obstacles to obtain the optimized conveying path. Specifically, a path re-planning algorithm such as the tangent method or the grid-based A* algorithm is used to generate a polyline path composed of multiple turning points, which avoids all path obstacle points and keeps a safe distance (such as more than 0.5 meters) from the obstacle boundary, ensuring that the conveying process will not collide. Then, the optimized conveying path is prioritized according to the moving distance, that is, the optimized conveying paths of all logs to be conveyed are sorted in ascending order of their total length. The shorter the path, the higher the priority of the log to be executed. This improves the overall operation efficiency and reduces the idle running time of the mechanical arm. Finally, an ordered log conveying path sequence is formed.For example, in an actual timber storage scenario, when a log marked as second-class is located at the starting coordinate (X=6.0, Y=4.0), the system looks up the second-class stacking area according to its grade and determines its storage end coordinate as (X=14.0, Y=7.0). Connecting the two points on the two-dimensional map forms an initial conveying path, but detection finds that the path needs to pass through a device access belt and interfere with a temporarily parked stacker. The system then identifies multiple path obstacle points and starts obstacle avoidance adjustment, re-planning a polyline path via the east side backup channel, including three turning points (6.0, 4.0)→(6.0, 6.0)→(12.0, 6.0)→(14.0, 7.0), forming an optimized conveying path. Comparing the distance with the paths of other logs, if the total length is 9.2 meters, it ranks third in the current batch, and is given a higher priority, and is finally included in the log conveying path queue, waiting for the mechanical arm to perform the transfer task.

[0042] In a specific scenario, the obstacle avoidance adjustment of the initial conveying path based on the path obstacle points to obtain the optimized conveying path includes: The size parameters of the path obstacle points are extracted to obtain obstacle size data, and the obstacle size data is used to delimit an obstacle exclusion zone; The initial conveying path is detected based on the obstacle exclusion zone to obtain a path intersection segment, and temporary turning points are set at the two end points of the path intersection segment; The temporary turning points and the non-intersection segment end points of the initial conveying path are connected by straight lines to obtain candidate obstacle avoidance paths, and the total lengths of the candidate obstacle avoidance paths are calculated and compared to select the shortest path as the optimized conveying path.

[0043] Specifically, the path obstacle points are subjected to size parameter extraction to obtain obstacle size data, and an obstacle exclusion zone is delimited based on the obstacle size data. This process first obtains the geometric information of the obstacles to which each path obstacle point detected on the initial conveying path belongs. These obstacles include fixed equipment, logs temporarily stacked, or other transport vehicles. The system determines the length and width of the obstacles based on the obstacle profiles pre-stored in the two-dimensional map of the warehouse area or in combination with real-time visual feedback. For example, the width of a stacker crane is 1.8 meters, and the length is 3.5 meters. Then, a safety distance (usually set to 0.5 meters) is expanded around the path obstacle point to form a rectangular exclusion zone surrounding the obstacle, ensuring that the robotic arm will not collide due to errors when running. Next, the initial conveying path is subjected to intersection point detection based on the obstacle exclusion zone to determine whether the originally planned straight path overlaps with the exclusion zone. If there is an overlap, two intersection points where the path enters and exits the boundary of the exclusion zone are extracted to form a path intersection segment located inside the exclusion zone. This path intersection segment is marked as an impassable section. Subsequently, temporary turning points are set at the two ends of the path intersection segment. Specifically, a unit distance (e.g., 0.6 meters) is offset along a direction perpendicular to the original path direction outside the entry point, and the direction is adjusted in combination with the surrounding passable area to make the temporary turning point fall within a safe passage. The same method is applied to the exit point side to generate a second temporary turning point. These two points will serve as key turning positions for the detour route. Then, the temporary turning points and the non-intersection segment end points of the initial conveying path are connected by straight lines. That is, the three straight lines between the start point and the first temporary turning point, the two temporary turning points, and the second temporary turning point and the end point are connected in sequence to form a complete candidate obstacle-avoiding path. This path completely avoids the obstacle exclusion zone and maintains path continuity. On this basis, the total length of the candidate obstacle-avoiding path is calculated and compared. The lengths of each straight line segment are added up using the Euclidean distance formula to obtain the total distance of the entire path. This value is compared with those of other possible detour schemes (e.g., alternative paths generated in different offset directions) to select the one with the shortest total length as the final optimized conveying path, ensuring the highest conveying efficiency. For example, in a real wood storage scenario, when the system detects that the initial conveying path needs to pass through an obstacle exclusion zone formed by a stationary forklift (with dimensions of 2.0m x 4.0m, and an expanded exclusion zone of 3.0m x 5.0m), it identifies that the path intersection segment is located in the middle of the exclusion zone. Then, one temporary turning point is set on each side of the intersection segment, offsetting to the east passage. After connection, a candidate obstacle-avoiding path in the shape of a "U" is generated, with a total length of 10.3 meters, which is superior to another alternative path that detours to the south (with a total length of 11.7 meters). Therefore, the former is selected as the optimized conveying path for the robotic arm to perform the transfer task.

[0044] In a specific scenario, the original wood is clamped and moved to the stacking coordinate area by the mechanical arm according to the original wood conveying path, comprising: The priority sorting information in the original wood conveying path is associated and bound with the grade identification in the corresponding original wood mark image to obtain a path-grade association table, and the path-grade association table is grouped according to the job batch to obtain a batch job list; Based on the batch job list, the mechanical arm is called to calibrate the position of the original wood corresponding to the first path in the batch job list to obtain a calibration clamping point, and the mechanical arm is moved to the calibration clamping point according to the original wood conveying path to obtain a to-be-clamped state; The mechanical arm in the to-be-clamped state is given a clamping instruction to clamp the fixed object, and the mechanical arm is driven based on the original wood conveying path to move the clamped fixed object to the corresponding stacking coordinate area.

[0045] Specifically, the priority sorting information in the log transportation path is associated with the grade identification in the corresponding log marking image to obtain a path-grade association table, and the path-grade association table is grouped according to the job batch to obtain a batch job list. This process first fuses the log transportation path data that has been planned with the corresponding log grade information. Each transportation path carries the starting point, ending point, obstacle point, path length and priority sorting information, while the grade identification (such as first grade and second grade) of each log in the log marking image is explicitly marked by color boxes and numbers. The system accurately matches the two through a unified log number or image index number to form a path-grade association table containing the fields of "path ID-starting point coordinates-ending point coordinates-priority-grade identification", ensuring that each transportation path has a unique corresponding relationship with the grade attribute of the corresponding log. Subsequently, according to the scheduling strategy of the current warehouse job, the entries in the path-grade association table are grouped according to the preset job batch capacity, for example, 5 logs are processed in each batch, the path entries with high priority are preferentially selected, and the grade distribution balance is also considered to avoid local congestion in the stacking area caused by processing only a single grade in a single batch. Finally, a structured batch job list is generated, which is arranged in execution order, and the first path corresponds to the highest priority task. Based on the batch job list, the mechanical arm is called, and the system sends a job start instruction to the mechanical arm control unit, loads all the path and grade information of the current batch, and enters the job preparation state. Subsequently, the log corresponding to the first path in the batch job list is positionally calibrated, that is, the real-time image of the area where the log is located is captured again through an industrial camera, and compared with the log marking image to detect whether the log has shifted or been blocked. If there is a deviation, the actual position coordinates are corrected through an image registration algorithm to obtain an accurate calibration gripping point, which is usually located near the center of gravity of the log and in a region where the gripping device can safely close. The mechanical arm is controlled to move to the calibration gripping point according to the log transportation path. The mechanical arm gradually approaches the target position along the initial segment in the original planned path according to the built-in motion control system and encoder feedback, and fine-tunes in the approaching process combined with visual servo technology to ensure that the end effector accurately aligns with the calibration gripping point. After reaching the calibration gripping point, the system determines that the mechanical arm is in a waiting-to-grip state. A gripping instruction is issued to the mechanical arm in the waiting-to-grip state, and the instruction is transmitted to the gripping device through the PLC control system to drive the hydraulic or electric clamping jaw to close and grip the log body. The gripping force sensor data is monitored in real time during the gripping process to prevent damage or insecure gripping.After the log is gripped, the robotic arm is driven to move the gripped object to the corresponding stacking coordinate area based on the log conveying path. The robotic arm moves segment by segment according to the turning point sequence in the optimized conveying path. During operation, the safety status of the surrounding environment is continuously monitored. If a sudden obstacle is encountered, an emergency stop or local replanning can be triggered. After reaching the destination, the log is smoothly placed at the designated coordinate point in the stacking coordinate area, the gripper is released to complete the stacking, and then it returns to the standby position or continues to execute the next batch of operations. For example, in a real timber storage scenario, when the system generates a batch operation list containing 5 logs, the first path is a primary log (numbered 02-01), which has the highest priority. After the system calls the robotic arm, it confirms that the position offset is 0.1 meters through camera re-measurement and corrects the calibration gripping point to (X=5.1, Y=3.9). The robotic arm accurately moves to this point to complete the gripping, and then follows the optimized path around the equipment area, finally moving it to the center coordinates of the primary palletizing area (X=10.0, Y=5.0) to complete the stacking. The entire process is automatically executed according to the path-level association table and the batch operation list, without the need for manual intervention.

[0046] The image recognition-based log storage, palletizing, and conveying method in the embodiments of the present invention has been described above. The image recognition-based log storage, palletizing, and conveying device in the embodiments of the present invention will be described below. Please refer to [link to relevant documentation]. Figure 2 One embodiment of the log storage, palletizing, and conveying device based on image recognition in this invention includes: The acquisition module 21 is used to acquire images of logs to be processed in the storage area using an industrial camera, and obtain log images. Extraction module 22 is used to extract log contour data from the log image and perform curvature detection on the log based on the log contour data to obtain log curvature parameters; The classification module 23 is used to classify logs according to their grade based on the log bending parameters, obtain log classification results, and mark logs of different grades in the log image based on the log classification results to generate a log marking image; The planning module 24 is used to perform spatial planning of the storage and palletizing area based on the log marking image to obtain the palletizing coordinate area, and to perform log transport path planning based on the palletizing coordinate area to obtain the log transport path. The transfer module 25 is used to pick up the logs to be processed and transfer them to the stacking coordinate area by means of a robotic arm, according to the log conveying path.

[0047] In this embodiment, the specific implementation of each unit in the above device embodiment is described in the above method embodiment, and will not be repeated here.

[0048] like Figure 3As shown, the embodiment of the present application provides a log storage stacking and conveying device based on image recognition, comprising: at least one processor; at least one memory for storing at least one program; When the at least one program is executed by the at least one processor, the at least one processor implements the above-mentioned log storage stacking and conveying method based on image recognition.

[0049] It can be seen that the content in the above method embodiments is applicable to the device embodiments, the device embodiments specifically implement the same functions as the above method embodiments, and achieve the same beneficial effects as the above method embodiments.

[0050] In addition, the embodiment of the present application also discloses a computer program product or a computer program, which is stored in a computer readable storage medium. The processor of the computer device can read the computer program from the computer readable storage medium, and the processor executes the computer program to enable the computer device to execute the above-mentioned log storage stacking and conveying method based on image recognition. Similarly, the content in the above method embodiments is applicable to the storage medium embodiments, the storage medium embodiments specifically implement the same functions as the above method embodiments, and achieve the same beneficial effects as the above method embodiments.

[0051] The above is a specific description of the preferred implementation of the present application, but the present application is not limited to the above-mentioned embodiments. Those skilled in the art can make various equivalent modifications or replacements without departing from the spirit of the present application, and these equivalent modifications or replacements are all included in the scope defined by the claims of the present application.

Claims

1. A log storage and stacking method based on image recognition, characterized by, The method comprises the following steps: image acquisition of the logs to be processed in the storage area by an industrial camera to obtain log images; extracting log contour data from the log images and detecting the bending degree of the logs based on the log contour data to obtain log bending parameters; classifying the logs based on the log bending parameters to obtain log classification results, and marking different grades of logs in the log images based on the log classification results to generate log marked images; based on the log marked images, the space of the storage stacking area is planned to obtain the stacking coordinate area, and the log conveying path is planned based on the stacking coordinate area to obtain the log conveying path; according to the log conveying path, the logs are clamped and moved to the stacking coordinate area by a mechanical arm.

2. The image recognition-based log storage and stacking transport method according to claim 1, characterized by, The image acquisition of the logs to be processed in the storage area by an industrial camera to obtain log images comprises: The logs to be processed in the storage area are partitioned and photographed in parallel by an industrial camera to obtain shooting images, and the shooting images are numbered and marked according to the area where the logs are located to obtain partitioned and numbered images; based on the partitioned and numbered images, the logs in each partition are subjected to synchronous contour segmentation to obtain single log images, and the single log images are subjected to area integrity verification to obtain log images.

3. The image recognition-based log storage and stacking transport method according to claim 1, characterized by, The extraction of log contour data from the log images and the detection of the bending degree of the logs based on the log contour data to obtain log bending parameters comprises: gray scale enhancement processing is performed on the single log images in the log images to obtain enhanced single log images, and log edge points in the enhanced single log images are identified; the log edge points are connected by contour fitting to generate an initial log contour, and redundant points in the initial log contour are removed to obtain log contour data; a straight line is drawn based on the two end points of the log contour data to obtain a log reference straight line, and a distance parameter of each pixel point on the log contour data to the log reference straight line is calculated; the maximum bending offset in the distance parameter is extracted, and a bending ratio is calculated based on the length of the log reference straight line and the maximum bending offset, and the bending ratio is taken as the log bending parameter.

4. The image recognition-based log storage and stacking transport method according to claim 1, characterized by, The marking of different grades of logs in the log images based on the log classification results to generate log marked images comprises: the grade identification in the log classification results is associated and matched with the single log images in the log images to obtain a grade-image correspondence table, and the single log images in the grade-image correspondence table are labeled with grade numbers to obtain numbered single log images; the contour boundaries of logs of different grades are determined based on the numbered single log images, and the contour boundaries are drawn using differential color boxes to obtain color box single log images; the color box single log images are batch spliced according to the partition layout of the log images to obtain partition marked images, and partition names and shooting time stamps are added to the partition marked images to obtain log marked images.

5. The image recognition-based log storage and stacking transport method according to claim 1, characterized by, The space planning of the storage stacking area based on the log marked images to obtain the stacking coordinate area comprises: The color frame single wood image in the original wood marking image is subjected to grade color recognition to obtain a grade color distribution, and the number of original woods of each grade is counted based on the grade color distribution; The storage space occupancy of the original wood of each grade is calculated based on the number, and the capacity allocation calculation of the storage space occupancy is performed according to the area of the storage area to obtain an allocated storage area; The allocated storage area is subjected to palletizing area segmentation by region division grid to obtain a hierarchical palletizing area, and the hierarchical palletizing area is subjected to position arrangement according to a preset grade priority to obtain an arranged palletizing area; The region coordinate points are pre-established based on the arranged palletizing area, and the region coordinate points are subjected to boundary range calibration to obtain a palletizing coordinate area.

6. The image recognition-based log storage and stacking transport method according to claim 1, characterized by, The original wood is subjected to conveying path planning based on the palletizing coordinate area to obtain an original wood conveying path, including: The starting point coordinate of a single original wood in the original wood marking image is located, and the storage end point coordinate of the corresponding grade original wood is determined based on the palletizing coordinate area; A straight line connection path is drawn in the storage area two-dimensional map based on the starting point coordinate and the storage end point coordinate to obtain an initial conveying path, and the initial conveying path is subjected to storage area obstacle detection to obtain a path obstacle point; The initial conveying path is adjusted based on the path obstacle point to obtain an optimized conveying path, and the optimized conveying path is subjected to priority sorting according to the moving distance to obtain the original wood conveying path.

7. The image recognition-based log storage and stacking transport method according to claim 6, characterized in that, According to the original wood conveying path, the original wood is clamped and moved to the palletizing coordinate area by the mechanical arm, including: The priority sorting information in the original wood conveying path is associated and bound with the grade mark in the corresponding original wood marking image to obtain a path-grade association table, and the path-grade association table is grouped according to the job batch to obtain a batch job list; The mechanical arm is called based on the batch job list, the position of the original wood corresponding to the first path in the batch job list is calibrated to obtain a calibration clamping point, and the mechanical arm is moved to the calibration clamping point according to the original wood conveying path to obtain a to-be-clamped state; The mechanical arm in the to-be-clamped state is given a clamping instruction to clamp the fixed object, and the mechanical arm is driven to move the clamped fixed object to the corresponding palletizing coordinate area based on the original wood conveying path.

8. A log storage and stacking conveyor based on image recognition, characterized by, It includes: The acquisition module is used for image acquisition of the original wood to be processed in the storage area by an industrial camera to obtain an original wood image; The extraction module is used for extracting original wood contour data from the original wood image, and the original wood is subjected to bending degree detection based on the original wood contour data to obtain an original wood bending parameter; The classification module is used for grade classification of the original wood based on the original wood bending parameter to obtain an original wood classification result, and the original wood of different grades is marked in the original wood image based on the original wood classification result to generate an original wood marking image; The planning module is used for space planning of the storage palletizing area based on the original wood marking image to obtain a palletizing coordinate area, and the original wood conveying path is planned based on the palletizing coordinate area. A transfer module is used to clamp and transfer the logs to be processed to the stacking coordinate area by a mechanical arm according to the log conveying path. 9.A computer device, comprising a memory and a processor, wherein the memory stores a computer program, and the computer device is configured to perform the method according to any one of claims 1-8. The computer program is executed by the processor to implement the steps of the method of any one of claims 1 to 7.

10. A computer-readable storage medium having stored thereon a computer program, characterized in that, The computer program is executed by the processor to implement the steps of the method of any one of claims 1 to 7.

Images