120 results about "Height map" patented technology

The invention discloses an ultra-thin glass defect detection system and positioning method based on line scanning and phase deflection, and the system comprises a line scanning camera module, a distortionless phase deflection detection module, a high-magnification microscope module and a central processing unit. The line scanning camera module is used for collecting a 2D grayscale image of the ultrathin glass; the distortionless phase deflection technology detection module is used for collecting a surface height map of the ultra-thin glass; the high-magnification microscope module is used for carrying out depth positioning on the interlayer defect judged by the central processing unit; the central processing unit is used for receiving the 2D gray level image and the surface height map, executing an image registration and defect classification algorithm and sending a depth positioning control instruction to the high-magnification microscope module, a multi-module cooperative work detection system is constructed, the detection efficiency and the detection precision are both considered, and data support is provided for production process improvement.

The invention belongs to the technical field of robot autonomous operation and reinforcement learning, and discloses an intelligent active perception and behavior interaction method in an unknown shielding scene. According to the method, a viewpoint planning strategy of an active sensing agent and a pushing strategy of a behavior interaction agent are closely fused, a 2.5 D occupancy height map is combined, and a pushing point screening mechanism of geometric features and task heuristic weights and a dense reward function are fused; the problems of insufficient perception information, low action efficiency and unstable strategy training in an unknown shielding environment are solved. According to the method, the pushing success rate, the environment entropy reduction efficiency and the strategy generalization ability are remarkably superior to those of a traditional method, efficient environment exploration can be achieved only through minimum invasive pushing of the scene, a good foundation is laid for tasks such as downstream grabbing and cleaning, and the method is suitable for robot autonomous operation of disordered shielding scenes such as bookshelves and storage cabinets.

An apparatus may be configured to perform a perception task based on features fused with height information. The apparatus configured may generate 3D sensor features from data from one or more sensors, generate a plurality of height maps from the 3D sensor features at a plurality of times, generate a height gradient map from the plurality of height maps, fuse the 3D sensor features and the height gradient map to generate height informed fused features, and perform the perception task using the height informed fused features.

This invention discloses a method for 3D mesh reconstruction of buildings from multispectral LiDAR point clouds, belonging to the field of multispectral LiDAR point cloud processing technology. The method includes: extracting building instances from the input point cloud using a multispectral LiDAR point cloud classification method; extracting geometric features from the height map obtained from the building point cloud, and then extracting spectral features from the spectral map using a spectral adaptive convolution module; inputting the data into a spectral-geometric fusion module to fuse the features of the spectral map and the geometric map to achieve accurate prediction of 2D corner points; projecting the 2D corner points onto the height index of the height map to obtain 3D corner points, and using an attention mechanism jointly guided by multispectral and geometric features to predict the boundary structure of the 3D corner points, generating a wireframe model; finally, achieving seamless and watertight 3D mesh reconstruction of the building under the guidance of the topological relationships contained in the wireframe model. This invention can effectively reduce artifacts and planar misalignment problems during the reconstruction process.

The invention provides a three-dimensional online boxing method considering a buffer area based on hierarchical reinforcement learning, and relates to the technical field of intelligent manufacturing, and the method comprises the steps: obtaining a size information list of a plurality of to-be-boxed boxes in the buffer area, and a three-dimensional height map of a current loading state in a container; extracting buffer area state feature vectors, and splicing the state feature vectors to obtain a global state feature vector; inputting the global state feature vector into a high-level strategy network in a trained hierarchical reinforcement learning network, outputting selection probability distribution of each box body in a buffer area, and selecting a target box body to be placed from the buffer area according to the probability distribution; and inputting the size information and the three-dimensional height map of the target box body into the trained low-layer strategy network, outputting probability distribution of each candidate placement position of the bottom surface of the container, generating an effective mask in combination with a preset physical constraint condition, and carrying out masking and renormalization processing on the probability distribution so as to determine a final placement position of the target box body.

The presently disclosed subject matter includes a computer system and a computer-implemented method of generating synthetic examination output images, including synthetic fault images and synthetic fault-free images. The synthetic images comprise artificially generated 3D defects. A machine learning model is trained for transforming examination output images to height maps. The height maps are modified to include certain 3D features, and a second machine learning model is used for transforming the height maps to modified examination output images to exhibit the 3D features.

The invention relates to the technical field of digital recognition and intelligent classification, in particular to an intelligent recognition and classification method for non-missing patterns, which comprises the following steps of: constructing a multi-dimensional embroidery pattern knowledge base, calculating a disparity map and a height map to obtain a stereoscopic strength coefficient, and calculating a stitch complexity index as a physical vector by combining a histogram of oriented gradients; meanwhile, the pattern recognition model is used for outputting the category probability of each independent pattern part as a semantic vector, then an embroidery method is determined, graph nodes are created based on physical and semantic vectors, a spatial relation edge and a cultural relation edge are established, then a multi-modal heterogeneous graph is generated, the graph is input into a graph neural network for message passing and global pooling, and the multi-modal heterogeneous graph is obtained. According to the method, a global feature vector is obtained, an embroidery theme is matched through a theme classifier, finally, the comprehensive credibility is calculated, and through multi-modal data fusion and heterogeneous graph analysis, the accuracy and automation level of pattern recognition are effectively improved, and the method is particularly suitable for intelligent classification of complex non-missing embroidery.

The application provides a surface feature-based anti-counterfeiting information generation method, device, equipment and medium, which comprises the following steps: firstly, a microstructure intensity map is obtained by extracting surface micro-relief features through a phase consistency model; a first height map is obtained by reconstructing an initial surface height field by using a MiDaS depth estimation network; a local displacement field caused by embedding a preset mark is quantified by using a RAFT optical flow model; an initial stress disturbance distribution is constructed by fusing the height field and the displacement field as physical input of a coupling prediction model, and then the surface morphology of a concrete sample after solidification is predicted; finally, the microstructure intensity map, the local displacement field and the first height map are taken as initial surface features at the sampling time, and the initial surface features and target surface features are encoded as anti-counterfeiting information, so that the reliability of sample identification is effectively improved.

A UAV control method, a device, a UAV, and a storage medium are provided. The method includes: obtaining a UAV's take-off position; obtaining a height map of an area around the take-off position; determining height information of a highest target in the area around the take-off position based on the height map; determining a relative height between the highest target and the take-off position based on the height information of the highest target; determining a restricted flight altitude of the UAV relative to the take-off position based on the relative height, and controlling the UAV based on the restricted flight altitude; where the restricted flight altitude is greater than or equal to the relative height. It thus can ensure flight safety while offering more reasonable and user-friendly flight flexibility and freedom.

To provide an object interference check technique using a height map for a machine tool and other applications.SOLUTION: The CAD model of the workpiece or other obstacle is converted into a height map that defines the obstacle height above the plane of the reference coordinate system for each cell of the grid. The tool center point path is then transformed into the same reference coordinate system. The interference check is performed by comparing the height coordinates of each transformed tool path check point with the values of the height map of the corresponding grid cell. The detected interference is handled by defining a new path point with an increased height coordinate to replace the interference critical path point. The interference of the entire tool can be checked by creating a bottom-to-top height map of the tool and comparing each pixel of the tool height map with the corresponding grid cell of the obstacle height map.SELECTED DRAWING: Figure 2

A method of convolutional neural network (CNN) based inversion of wave radar images and use thereof pertain to the field of sea-state monitoring, in which a CNN autoencoder is employed. A training dataset produced by a simulation tool is input to the CNN autoencoder for supervised training, with X-band radar images of an inversion region being taken as input to the training dataset and with corresponding ocean surface wave height maps being taken as output from the training dataset, thereby deriving a model defining a mapping of them. The CNN autoencoder includes an encoder, fully connected layers and a decoder. The encoder includes five convolutional layers and five max-pooling layers, which are connected alternately. There are two symmetric fully connected layers, and the decoder includes five deconvolutional layers. A CNN-based deep learning algorithm is used to provide good reconstruction of the nonlinear components. The autoencoder technology is employed, which can directly reconstruct real-time ocean surface wave height maps from X-band radar wave images with minor errors, which provide more comprehensive sea-state information.

To improve the measurement accuracy of a pin even when the tip shape of the pin is different.SOLUTION: A setting part 31c of a setting support device 1 of a measuring instrument has a tool for measuring a predetermined measuring object as a measuring item, receives designation of a measuring area and designation of an object area for specifying the measuring object in the measuring area in response to selection of the tool, and sets the tool including the measuring area and a filter parameter for measuring the measuring object according to the designation of the measuring area and a size of the object area and / or a representative height in the object area. The symbol corresponding to the target region is displayed on the height image, the measurement target corresponding to the filter parameter in the measurement region is specified based on the setting of the tool, and the measurement of the tool is executed based on the measurement value of the specified measurement target.SELECTED DRAWING: Figure 3

The invention provides an anti-fake information generation method and device based on surface features, equipment and a medium. The method comprises the steps that firstly, surface micro-fluctuation features are extracted through a phase consistency model to obtain a microstructure strength diagram; reconstructing the initial surface height field by using a MiDaS depth estimation network to obtain a first height map, and quantifying a local displacement field caused by embedding of a preset identifier through an RAFT optical flow model; the height field and the displacement field are fused to construct initial stress disturbance distribution to serve as physical input of a coupling prediction model, and then the surface appearance of the concrete sample after solidification is predicted; and finally, the microstructure strength diagram, the local displacement field and the first height diagram are used as initial surface features during sampling, and the initial surface features and the target surface features are coded into anti-counterfeiting information, so that the reliability of sample identification is effectively improved.

The invention discloses a luggage dynamic stacking method based on point cloud segmentation and reinforcement learning, a robot device and application thereof, and is suitable for complex scenes such as airports. The robot device comprises a three-dimensional sensing module, a calculation processing module, a robot execution module and a ULD loading module, and real-time data interaction is achieved through the industrial Ethernet. The core of the method is'perception-decision-execution 'closed loop: a three-dimensional sensing module captures three-dimensional point cloud data of an operation space, and a structured scene representation is generated through point cloud segmentation, OBB calculation and multi-channel 2D height map construction; based on a reinforcement learning model of an Actor-Critic framework, an optimal 6D placement pose is output through offline training and online execution; and the robot execution module accurately completes luggage grabbing and placing. According to the method, traditional off-line planning is abandoned, real-time sensing replaces ideal model dependence, reinforcement learning replaces fixed rules, the method can dynamically adapt to changes such as luggage deformation and displacement, and stacking robustness, space utilization rate and compatibility are improved.

The invention relates to a displacement imaging system suitable for semiconductor IGBT power module detection, which comprises a camera, a lens, a coaxial illumination module, an execution mechanism, an image processing module and a control module, and is characterized in that the execution mechanism drives the lens to reciprocate up and down in a direction vertical to an IGBT module; the control module controls the execution mechanism to drive the lens to move up and down and controls the camera to shoot images. And the image processing module is used for calculating the sharpness of different image areas of each image, performing height mapping according to the height of the sharpness of the image areas to obtain a height map, and using the gray value of the clearest image in the area as the gray value of the full-clear image in the same area to finally obtain the final full-clear image. According to the method, the height information of each point on the surface of the IGBT module can be accurately obtained to generate the height map, so that quantitative detection of three-dimensional shape parameters such as the height of the aluminum wire is realized, and the visibility and detectability of defects are greatly improved.

The application relates to the technical field of robot grabbing, and particularly discloses a pushing-grabbing reward design method of a grabbing affordance improved Q network, which comprises the following steps: collecting RGB-D image information of a working scene, obtaining a color height image and a depth height image through projection transformation, inputting the color height image and the depth height image into a pushing network, a grabbing network and a pre-trained grabbing affordance network respectively, sequentially generating a pixel-level cumulative reward expectation value Q push , a Q grasp and an affordance value A grasp ; selecting a grabbing affordance value calculated by taking the proportion of the minimum pixel value of a grabbing object in the scene as a set threshold, further judging whether the number of pixels with an affordance value not less than 1 exceeds the set threshold, and selecting a grabbing operation or executing a pushing action, so that the rationality of grabbing and pushing behavior evaluation is improved.

The invention discloses an air luggage online stacking method and device based on deep reinforcement learning, and belongs to the technical field of air luggage stacking. The method comprises the following steps: training a dual-depth Q learning network by using a plurality of training samples; the training mode comprises the steps that a first luggage queue and a height map of a training sample are input into a network, the network constructs a state space based on the height map and the first luggage queue, to-be-stacked luggage is selected from the first luggage queue one by one, and for the to-be-stacked luggage selected each time, the following steps are executed: the action space of the to-be-stacked luggage is determined, and the action space of the to-be-stacked luggage is determined; updating the state space by using the action space; and the stacking space utilization rate is determined, and the stacking mode is output based on the maximum space utilization rate after iterative updating. According to the method, the space utilization rate of luggage packing can be greatly improved, real-time decision making can be carried out in a complex dynamic online environment, the method can adapt to heterogeneous luggage of different sizes, shapes and packing quantities, and robustness is guaranteed.

The present disclosure relates to a method for prediction of process defects performed by at least one processor, the method including receiving design data including a plurality of layout patterns corresponding to a plurality of layers of a semiconductor chip, dividing the design data into tile regions, and generating a plurality of height maps corresponding to a plurality of tile regions, and determining a presence of process defects associated with the plurality of tile regions based on the plurality of height maps.

This invention discloses a method, system, and storage medium for LOD2 building reconstruction from single-view orthophotos. The method inputs a single-view orthophoto remote sensing image into a shared backbone network to generate a footprint mask probability map; constructs a first modulation factor, performs element-wise gating on the multi-scale feature map, and inputs it into a height decoder to output a height field; normalizes the height field to obtain a normalized height map and constructs a second modulation factor; fuses the second modulation factor with the first modulation factor to form a composite modulation factor, performs dual element-wise gating on the multi-scale feature map, and inputs it into a structure line decoder to output a distance field and orientation field; recovers the roof structure line based on the distance field and orientation field; and constructs the roof surface, walls, and ground by combining the roof structure line, height field, and footprint boundary to generate a LOD2 3D model of the building. This invention can automatically generate a topologically complete and geometrically accurate LOD2 3D model relying solely on a single orthophoto.

The invention discloses a bulk material vehicle intelligent AI loading method and system based on laser radar point cloud and intelligent. A system collects loading area point cloud, fitting is carried out to obtain a container boundary polygon and a loading surface height map, and a target filling field is constructed; a discharging influence function is established, the discharging pose, the pose and the bucket / gate opening degree-time curve are mapped into height increments, and a height map after discharging is predicted; carrying out cutting convolution on the area outside the container boundary to obtain an overflow index and a container edge risk zone; partition point selection-drop point rollback-opening amplitude limiting / duration shortening are executed through a two-stage parameter solver, and spilling is restrained while a loading target is met. And after unloading, recursive least square or Kalman filtering is adopted for online updating based on height map difference, cross-material and cross-vehicle type self-adaptive loading control is achieved, spill and rework are reduced, and filling consistency and operation efficiency are improved.

The present application relates to the technical field of image detection, and discloses a card surface topography defect detection method and system based on a relief effect, which comprises the following steps: double-frequency structured light adaptive projection of a laminated card, relief three-dimensional height map reconstruction based on phase deflection, automatic segmentation of a relief area based on a variational level set, analysis of relief profile curvature continuity and defect detection, and relief depth statistical process control. The present application eliminates mirror reflection interference of a laminated layer through a double-frequency phase shift algorithm, and the measurement accuracy of the relief depth reaches ±5 microns, and the detection accuracy reaches 99.5%.

The application discloses a virtual terrain landscape generation method and system, an electronic device and a storage medium, and relates to the technical field of computers. The method comprises the following steps: generating original block models with different inner surface detail levels; each original block model is square in shape, and the number of surfaces of the outermost layer is the same; the number of surfaces of the inner surfaces of the original block models with different inner surface detail levels is different; a central processing unit generates a terrain height map, a target ground map array, a target normal map array, a ground control map and a void model; a graphics processing unit performs seamless connection on the original block models with different inner surface detail levels according to the terrain height map to obtain a terrain model; a ground map is rendered according to the ground control map, the target ground map array and the target normal map array to obtain a ground map model; and the ground map model, the terrain model and the void model are combined to obtain a target virtual terrain landscape. The performance consumption can be reduced in the process of generating the virtual terrain landscape.

This application proposes a method for identifying drivable road areas based on binocular vision. The method includes: acquiring left and right images of the road using a binocular camera; extracting and matching features from the left and right images based on a spatial attention mechanism to obtain an attention matrix and image features, and generating a disparity map based on the attention matrix and image features; performing a first coordinate transformation on each pixel in the disparity map to obtain a corresponding depth map, and performing a second coordinate transformation on each pixel in the depth map to obtain a corresponding height map; and identifying the drivable area of ​​the road based on the height map. The technical solution of this application is applicable not only to road surface information perception on paved roads but also to road surface information perception on unpaved roads, and the perception result is the drivable area ahead of the road, which facilitates the autonomous driving decision-making module to make decisions directly based on the drivable area.

This invention discloses an automatic frequency-elevation map measurement method and system based on residual networks, enabling rapid and automatic extraction of key ionospheric parameters from frequency-elevation maps acquired by digital altimeters. The method first parses the header information and echo intensity data from the original binary file of the altimeter to generate a frequency-elevation map image. Then, a small number of samples are manually labeled to establish a training set with critical frequencies such as foF2 and h'F2 as ground truth. Based on this, an end-to-end model including a backbone residual network, pooling layers, and a feature extraction head is built. This model is trained to convergence through backpropagation and further fine-tuned using expanded sample data to obtain an engineering-deployable model. Finally, the frequency-elevation map to be processed is input into the model, and the predicted values ​​of the target parameters are output and written to the results file. This method reduces the workload of manual interpretation and verification and improves the measurement accuracy of irregular and disturbed frequency-elevation maps.

Embodiments of the present application disclose a path generation method and device for a virtual scene, computer equipment and a computer readable storage medium. According to the present application, a starting point and an ending point are selected in a virtual scene height map, the starting point is taken as a path node, the slope value between the path node and each adjacent point is determined according to the height information of the path node and each adjacent point in the virtual scene height map, the movement consumption value generated by the path node through each adjacent point to reach the ending point is determined according to the slope value and the distance value between each adjacent point and the path node and the ending point, the next path node of the path node is determined from at least one adjacent point based on the movement consumption value, the next path node is taken as a new path node, the next path node of the new path node is determined again until the next path node coincides with the ending point, and a target path is obtained. In this way, the path generation efficiency in the virtual scene can be improved.

The invention discloses a low-error surveying and mapping unmanned aerial vehicle and a surveying and mapping method, relates to the technical field of distance measurement, and can solve the technical problems of low data processing precision and poor efficiency in a surveying and mapping process due to the fact that real topographic relief and unmanned aerial vehicle bumping errors cannot be effectively distinguished under complex topographic conditions. Comprising the steps that a ground height data set obtained by measuring a target area through a surveying and mapping unmanned aerial vehicle is acquired, and the ground height data set comprises ground height data of multiple measurement points; screening out a plurality of alternative error data from the ground height data set, wherein the alternative error data are measurement point data deviating from a normal fluctuation range; multiple pieces of target error data are identified from the multiple pieces of alternative error data, wherein the target error data are error data caused by jolting of the surveying and mapping unmanned aerial vehicle in the multiple pieces of alternative error data; performing data correction on the target error data to obtain a corrected ground height data set; and determining a ground height map according to the corrected ground height data set.

Methods and systems for high throughput and high resolution measurement of surface topography of substrates employed in semiconductor device manufacturing are described herein. A beam steering device scans an illumination beam along a beam scan path incident on a substrate surface. In addition, a specimen positioning system moves the substrate with respect to the illumination beam. In some embodiments, a hypercentric focusing lens in the illumination beam path between the beam steering device and the substrate eliminates the need for an optical element in the optical path between the surface under measurement and a position sensitive detector in the optical path of the reflected beam. A computing system generates a map of wafer tilt based on the detected position of the reflected beam at the detector. In a further aspect, a high resolution surface height map is generated based on the measured surface tilt map.

The application relates to the technical field of image processing, and particularly discloses a floor engineering acceptance system based on image acquisition, which is used for solving the problems that in floor acceptance images, measuring points and directions are random, near edges and holes are interfered, reconstruction noise and scales are not unified, virtual reading is easy to cross holes and borders, calibers are inconsistent, it is difficult to align with on-site reading data one by one, and traceable review is lacked, and the system comprises an imaging acquisition unit, a three-dimensional reconstruction calibration unit, a reference surface fitting and detrending unit, a height map gridding unit, a wavelet scale management unit, a dimension mapping unit, a boundary and hole processing unit, a candidate measuring area generation unit, a virtual reading determination unit and a data storage and output unit; through automatic reading consistent with on-site reading caliber, the floor acceptance image is prevented from crossing holes and borders, near edge interference is inhibited, traceable measurement alignment is established, error boundaries and reserved digits are determined, and unified statistical caliber and version records are established.

The invention relates to the technical field of image processing and pattern recognition, and discloses a self-adhesive film adhesive force evaluation method based on pattern recognition. A contact area height map is obtained through three-dimensional measurement, and zero offset is achieved; projecting the height matrix column mean value to generate a one-dimensional elevation curve; discrete curvatures are extracted through three-point difference, and the total amount is obtained through accumulation; accumulating coefficient squares to obtain total energy by using a self-defined non-orthogonal spectral basis expansion curve; solving the mean value, the standard deviation, the skewness and the kurtosis of the energy sequence; frequency spectrum intensity is generated column by column according to the coefficient and the spectrum basis, a mean value and a standard deviation are taken to identify a salient region, and the average intensity is solved; constructing an adhesive force evaluation function according to the total curvature, the total energy, the skewness and the kurtosis; and forming a normalized adhesion index by combining the evaluation value and the average intensity of the salient region, and grading according to a threshold value.