6 results about "Contour mapping" patented technology

The invention discloses a counterweight type robot polishing automatic clamping system which comprises an intelligent workpiece adaptability automatic clamping module and a control system. The intelligent workpiece adaptability automatic clamping module comprises a visual recognition unit, an execution mechanism, a clamping assembly and a force feedback unit. The visual identification unit is used for collecting image information of a to-be-polished workpiece and analyzing and determining the shape, size and surface feature parameters of the workpiece; the control system is based on the shape, size and surface characteristic parameters of a workpiece. The invention relates to the technical field of robot automation. According to the counterweight type robot polishing automatic clamping system, through multi-angle image collection and processing of the visual recognition unit, the system can accurately obtain the shape, size and surface characteristics of a workpiece, automatically analyze and generate a three-dimensional contour mapping graph, provide a basis for subsequent clamping path planning and clamping point optimization, have high adaptability and are suitable for large-scale popularization and application. And diversified workpieces can be handled.

The application provides a UAV naked-eye 3D two-three-dimensional digital mapping method, proposes three-dimensional contour mapping based on point clouds, multi-storey building digital mapping based on real scene models, and digital orthographic image-based digital mapping of ground objects and landforms, realizes rapid mapping of the terrain, ground objects and landforms in the construction area, and the mapping accuracy is better than 5cm, meets the large-scale digital mapping requirements, improves the engineering project construction deployment efficiency, and is suitable for three-dimensional terrain mapping of all construction areas such as railways, highways and municipal engineering. It belongs to the field of surveying and mapping.

This invention provides a method, server, and storage medium for predicting latent anomalies in copper-clad laminate (CCL) lamination based on contour pattern recognition. By acquiring sampling sequences of CCL edge contours and a reference contour sequence, a cross-scale contour mapping state field is generated through dual-dimensional curvature and fractal encoding. Gradient flow decomposition is performed on the mapping field to separate geometric constraint components and instability components, forming an intrinsic contour pattern distribution manifold. Trend mutual induction measurement is performed on the perturbation components corresponding to latent anomalies such as interlayer slip and uneven resin wetting in multiple batches of manifolds, generating an evolution trend map. When the two types of perturbation components in the map synchronously and monotonically increase within the anomaly response time window and the slope convergence exponent exceeds a threshold, a prediction signal is generated. This invention achieves early coupled prediction of latent anomalies in CCL lamination, improving quality monitoring capabilities and the timeliness of process control response.

The invention relates to the technical field of intelligent traffic facility monitoring, in particular to a barrier gate fault detection method and system based on image recognition and instruction verification, and the method comprises the following steps: obtaining image frame sequence equalization image brightness, extracting a contour mapping coordinate judgment state, comparing control time, extracting an abnormal frame, tracking a serial number analysis direction, and eliminating interference. And position change is extracted to compare control missing, and a barrier gate abnormal action fault section number set is obtained. According to the method, through corresponding analysis of image frame states and control instruction time periods, state changes caused by non-instructions are separated, frame sequence numbers and direction consistency judgment are combined, interference frame segments are eliminated, position offset starting nodes are extracted, image sections with concentrated action deviation are positioned, and the image frame states and the control instruction time periods are compared in an image mapping and state direction comparison mode. Abnormal action recognition which is continuous in offset and lacks control signal support is completed, the tracking capability of the state change process is enhanced, and the mining range of action abnormity and the positioning precision of a fault section are improved.

The application discloses a contour mapping method, device, equipment and storage medium, selects an initial vertex from each first vertex of a three-dimensional closed contour, breaks at the initial vertex to obtain an initial open contour, and the initial vertex becomes two first vertices; maps the first vertex on a plane to obtain each second vertex, groups each second vertex to form a first open contour, and the two second vertices obtained after the initial vertex is changed are not connected; closes the first open contour to obtain a first closed contour; judges whether the first closed contour contains at least two closed areas; if yes, a target area is determined, intersection elimination processing is performed on the target area, a new first closed contour is obtained, and the step of judging whether the first closed contour contains at least two closed areas is returned to be executed; and if no, the first closed contour is taken as a target mapping contour. The target mapping contour obtained by the scheme is more accurate and closer to the real shape of the three-dimensional closed contour.

The invention discloses a medical image lesion area identification method and system based on deep learning, and the method comprises the steps: extracting a multi-modal feature sequence from original image data collected by lung CT equipment, and obtaining a denoised feature image set; determining initial edge contour mapping according to the pixel distribution difference in the de-noised feature map set; obtaining regional connected components in the preliminary edge contour mapping, and obtaining an enhanced connected region set by calculating a gray similarity matrix between the components; for the enhanced connected region set, performing semantic segmentation operation on each region by adopting a U-Net model to obtain a fine segmentation mask; extracting a boundary coordinate sequence from the fine segmentation mask, and obtaining a positioning coordinate group; obtaining a heterogeneity index in the positioning coordinate group to obtain a robust positioning result; and according to lesion boundary information in the robust positioning result, generating a diagnosis report superposition layer, and superposing the diagnosis report superposition layer on the original image data to obtain final visual lesion identification output.