19 results about "3d image processing" patented technology

A system and method of three-dimensional image processing. comprising the steps of pre-processing at least one set of raw images each including a plurality of two-dimensional source images. wherein each of the two-dimensional source image represents a cross-sectional view of a three-dimensional object at different positions along an axis in a three-dimensional space: and constructing a three-dimensional image representing the three-dimensional object by integrating 3D mesh points extracted from the at least one set of raw images being pre-processed: wherein the three-dimensional image is readable by a first image viewer arrange to render and to facilitate manipulation of the three-dimensional image.

The invention discloses a binocular vision naked eye 3D image processing method, device and system, and particularly relates to the technical field of naked eye 3D. According to the method, the foreground and background areas are defined, a parameter extraction and merit figure quantification calculation system covering parallax, scenes and splicing dimensions is constructed, and the multi-scene preset threshold and the weight factor are combined, so that comprehensive and accurate evaluation of the naked eye 3D image quality is realized, the problem of one-sided evaluation of a traditional single index is solved, and the evaluation efficiency is improved. And the core crux influencing the three-dimensional effect is accurately positioned.

The invention relates to the technical field of three-dimensional image processing, in particular to a distributed parallel processing and loading method for large-scale scene three-dimensional model data, which comprises the following steps of: acquiring a plurality of large-scale scene three-dimensional model patch coordinate sets, normal vector sets and material identification sets, calculating an adjacent matrix and judging a spatial aggregation region; the method comprises the following steps: segmenting and generating regional data subsets, calculating patch quantity and texture size matching calculation node capability, constructing a distribution path table, extracting a continuous triangular patch chain and reconstructing compressed data, calculating a loading priority in combination with a view angle, and loading and caching data according to scheduling. The spatial aggregation degree of adjacent patches is calculated, spatial regions are divided by combining material change points, the computing power and bandwidth capacity of nodes are dynamically matched and computed according to the patch number and the texture size, and the screen space intersection area and the viewpoint distance are computed by combining the user view angle to adjust the loading priority. And the visual angle-oriented visual scheduling response efficiency is improved.

The techniques described herein relate to methods and systems for three-dimensional (3D) image processing using deep learning model pre-trained with two-dimensional (2D) images. The techniques include transforming a 3D representation to a 2D map, which can be input to the model. The output of the model could be a defect segmentation mask, or a probability of the input belonging to a given category. The input 2D image can result from a particular transformation configuration. The techniques described herein provide for an effective and efficient method of identifying a transformation type for applications such as 3D data-based classification, anomaly detection and segmentation. The techniques described herein perform statistical analysis over a set of training samples to identify a transformation type. The statistical analysis can include histograms of pixel values.

The application provides an endoscope, an endoscope assembly and a surgical microscope system. Two or more light beams collected by two or more light collecting modules from different angles of an observation target are combined into coaxial propagation light beams by a beam combination module. An image sensor is arranged on an image side after beam combination to image the light beams collected from different angles of the observation target, so as to obtain stereoscopic image information of the endoscope observation target. The present application changes double optical path imaging into single optical path imaging by arranging the beam combination module, reduces part of the optical path, reduces the number of optical elements, reduces the cost of the endoscope, realizes miniaturization of the endoscope, and calibrates three-dimensional amplitude. Only the optical elements need to change the light propagation path physically, without the need for complex 3D image processing algorithms, and the number of optical elements that need to be precisely assembled and calibrated can be reduced, so as to reduce the complexity of imaging.

This application discloses a multi-scale structure-aware optical field depth estimation method and system, relating to the technical field of 3D image processing. The method includes: constructing a candidate neighborhood micro-image set; constructing and scoring a consistency scoring function for each candidate view in the candidate neighborhood micro-image set, selecting the candidate views with the highest scores to obtain an adaptive neighborhood micro-image set; projecting reference pixels onto each neighboring micro-image in the adaptive neighborhood micro-image set, calculating the matching cost, and performing robust aggregation based on fuzzy difference weights; performing structure-aware propagation repair, reconstructing the matching window and repairing the cost propagation chain; iteratively updating the disparity plane, outputting a disparity map for each reference micro-image, fusing the disparity maps into a virtual depth map, and mapping the virtual depth to a measurable object distance to obtain the final depth map. This application can improve the accuracy and completeness of multi-focal-length optical field depth estimation.

The invention provides a chip bonding wire three-dimensional reconstruction method and system based on three-color mask single-camera multi-view imaging, and relates to the technical field of three-dimensional image processing. The method comprises the following steps: constructing a single-camera optical imaging system containing a three-color mask, and carrying out brightness and geometric calibration; acquiring an original image fused with the visual angle information through single exposure; generating a three-view image through color separation and interpolation resampling; after pixel-level brightness and color correction is carried out on the image, multiple groups of optical flow fields are calculated and fused to obtain a unified displacement vector field; spatial back projection is carried out by combining parallax information and the calibration matrix to construct dense point cloud, and a three-dimensional grid model is obtained through outlier elimination and curved surface reconstruction; and finally finishing texture mapping and outputting a three-dimensional reconstruction result with real color information. Under the condition that multiple cameras and a scanning mechanism are not needed, high-precision, real-time and non-contact three-dimensional reconstruction of a complex chip lead structure is achieved, and the method is suitable for application scenes such as chip packaging quality online detection.

The invention provides a low-calculation-requirement 3D image processing method and system. The method specifically comprises the steps that collected left-eye and right-eye images and image display and calibration information actually to be obtained are obtained; constructing a coordinate transformation matrix, and performing translation zooming processing on the left eye image and the right eye image; constructing a common internal reference matrix of the left-eye camera and the right-eye camera, performing epipolar correction on coordinates of the left-eye image and the right-eye image after translation and zooming, and performing distortion correction on the coordinates of the left-eye image and the right-eye image after epipolar correction; performing first interpolation to obtain displayable left and right eye images; calculating left and right disparity maps based on the interpolated left and right eye images, and calculating a depth map based on the left and right disparity maps to obtain depth information; reconstructing a 3D point cloud based on the depth information and the left and right eye image coordinates after distortion correction; and rotating the 3D point cloud, regenerating left-eye and right-eye 2D image coordinates, performing second interpolation, and finally outputting the image. The method has the advantages of high precision and low delay.

The application discloses a binocular vision naked-eye 3D image processing method, device and system, and particularly relates to the naked-eye 3D technical field; the application defines foreground and background areas, constructs a parameter extraction and optimal value quantization calculation system covering parallax, scene and splicing dimensions, and combines preset threshold values and weight factors in multiple scenes to realize comprehensive and accurate evaluation of naked-eye 3D image quality, solve the problem of one-sidedness of traditional single-index evaluation, and accurately locate the core crux affecting the stereoscopic effect.

The techniques described herein relate to methods and systems for three-dimensional (3D) image processing using deep learning model pre-trained with two-dimensional (2D) images. The techniques include transforming a 3D representation to a 2D map, which can be input to the model. The output of the model could be a defect segmentation mask, or a probability of the input belonging to a given category. The input 2D image can result from a particular transformation configuration. The techniques described herein provide for an effective and efficient method of identifying a transformation type for applications such as 3D data-based classification, anomaly detection and segmentation. The techniques described herein perform statistical analysis over a set of training samples to identify a transformation type. The statistical analysis can include histograms of pixel values.

The invention provides an endoscope, an endoscope assembly and an operating microscope system. The light beams collected by two or more than two light receiving modules from different angles of an observation target are combined into coaxially propagating light beams through a beam combining module; and an image sensor is arranged at the image space after beam combination to image the light beams collected from different angles of the observation target, so that the three-dimensional image information of the observation target of the endoscope is obtained. According to the scheme, through the arrangement of the beam combining module, dual-optical-path imaging is changed into single-optical-path imaging, and part of optical paths are reduced, so that the number of optical elements is reduced, the cost of the endoscope is reduced, and the microminiaturization of the endoscope is realized; according to the three-dimensional amplitude calibration, only the optical elements need to physically change the light propagation path, a complex 3D image processing algorithm is not needed, the number of the optical elements needing to be precisely assembled and calibrated can be reduced, and therefore the imaging complexity can be reduced.

The 3D scanner and the 3D scanning system comprise a light source, an image acquisition module, a first communication assembly, a second communication assembly and a 3D image processing module. The light source projects a light signal to a target object, and the image acquisition module collects the light signal reflected by the target object and generates an image sequence. The first communication assembly receives an image sequence of the image acquisition module and transmits the image sequence to the second communication assembly, and the 3D image processing module receives the image sequence from the second communication assembly and performs 3D reconstruction. According to the utility model, the 3D image processing module is integrated into the 3D scanner, and the software and hardware compatibility test of the 3D image processing module is completed before the 3D scanner leaves the factory, so that the compatibility problem occurring when a user configures a computer by himself / herself is avoided. In addition, the device has the advantages of wide industrial application and simplicity and convenience in operation.

The invention provides a bus-controllable video parallel processing method and device and a server, and relates to the technical field of image processing, and the method comprises the steps: collecting to-be-processed video data, and determining a target segmentation line parameter based on the storage data bit width of a storage device and the resolution information of the to-be-processed video data; according to the target segmentation line parameter, performing image segmentation processing and mirror image edge filling processing on each frame of input image in the to-be-processed video data, determining a target left segmentation image and a target right segmentation image after segmentation, and respectively storing the target left segmentation image and the target right segmentation image in a left memory and a right memory; and performing 3D image processing on the stored left target segmentation image and the right target segmentation image, and when the 3D image processing is completed, extracting and splicing the left target segmentation image and the right target segmentation image into a video stream for display. According to the method, the memory overhead and the processor burden during video processing can be remarkably reduced while the video image quality is guaranteed.

The present invention provides a data reading / writing method and system in 3D image processing, a storage medium, and a terminal. The method comprises the following steps: for each sub-image, storing processing data of the sub-image to a cyclic buffer area; after the sub-image is processed, reserving data of an overlapped part required by the next sub-image in a cyclic buffer area; and dividing a multi-layer network of an image processing algorithm into at least two stages, so that data in adjacent layers of each stage is exchanged only by means of the buffer, without using DDR interaction. On the basis of 3D vertical sliding technology and a circular buffer area, the data reading / writing method and system in 3D image processing, the storage medium, and the terminal in the present invention can significantly improve buffer utilization in 3D image processing in a situation of a limited buffer, reduce processing on an overlapped part, and thus alleviate bandwidth consumption and a reading / writing delay problem in image processing as a whole.

The invention relates to the field of 3D image processing, in particular to an augmented reality-based three-dimensional model data processing system, which comprises a data acquisition module, a grid classification simplification module, a three-dimensional model simplification module and a display module, according to the three factors of the change of the normal vector of the triangular mesh surface, the geometric structure and the geometric structure distribution, subspace structure complexity is constructed; obtaining the complexity of subspace types; obtaining subspace feature relevance; obtaining the complexity of the subspace; completing the division of the subspaces; and simplifying the grid vertex of the three-dimensional model by utilizing a clustering algorithm according to a division result. Therefore, the simplification processing of the three-dimensional model data is realized, the problem of shape deformation or data loss of the three-dimensional model in the simplification process is greatly avoided, and the error between the simplified model and the original model is reduced.

The present invention discloses a method for extracting characters from curved surface parts based on 3D point clouds, which belongs to the fields of industrial manufacturing and 3D image processing. Based on 3D point cloud data analysis, the present invention innovatively applies 3D image processing technology to industrial production sites, completing the task of extracting characters from curved surface parts. The invention first uses a 3D camera to collect the shape data of the curved surface parts and saves the data in a 3D point cloud format; then uses a character positioning algorithm to find the exact position of the curved surface characters, and extracts the point cloud data of the character area separately; then performs surface correction on the extracted character data one by one; finally, uses the characteristics of the normal vector of the 3D point cloud to extract the characters on the surface. The method described in the present invention can accurately locate the position of the characters on the curved surface parts and realize the extraction of the curved surface characters. It still has good robustness under conditions such as non-uniform illumination and small color difference between the character area and the non-character area, solving the problems of difficult extraction of characters from curved surface parts and poor extraction effect.

The invention relates to the technical field of real-time three-dimensional image processing, and provides a method and a device for improving the visual fidelity of an image. According to the method, on the basis of the assumption that original pixels are located between two adjacent surfaces in a source image, the positions where the original pixels are located are classified, and different intersection filters are used for processing the original pixels meeting corresponding classification standards; after the intersection filter responds, an amplification coefficient corresponding to the decomposed source image features in at least one dimension is determined according to the processing logic of the intersection filter, original pixels are amplified based on the amplification coefficient, and a good enough approximate estimate of the amplified target image is obtained; according to the image super-resolution algorithm, the specific parameterization of the image is realized, the support of a filter is reduced to the maximum extent, the image quality and the calculation complexity are balanced to the maximum extent in the limited fixed function hardware prediction, and the problems that the image super-resolution algorithm in the prior art is not suitable for hardware and is poor in effect are solved.

The method of the present invention discloses a cross-modal interactive RGB-D image salient region detection method. The method of the present invention first extracts RGB features and depth features from the color image and depth image respectively using the VGG16 network; then, in the RGB-D interaction module, the RGB features and depth features are mutually corrected and edge details are enhanced; in the depth weight allocation module, the fusion weights of the processed multiple features are calculated based on the depth features; finally, the fused features are refined in the attention update module. The method of the present invention links RGB with depth, and enhances the target area by supplementing rich boundary information; the weights are calculated in the multi-scale depth weight module, and the degree of influence on the salient area is determined according to the quality of the depth map, thereby eliminating the negative impact of low-quality depth maps and improving the overall performance of the algorithm. The method of the present invention can be directly applied in engineering fields such as 3D image processing, 3D quality assessment, and object recognition.