CN109658522B - Simplified method for temporary label merging processing of three-dimensional binary image connected body label - Google Patents
Simplified method for temporary label merging processing of three-dimensional binary image connected body label Download PDFInfo
- Publication number
- CN109658522B CN109658522B CN201811427359.1A CN201811427359A CN109658522B CN 109658522 B CN109658522 B CN 109658522B CN 201811427359 A CN201811427359 A CN 201811427359A CN 109658522 B CN109658522 B CN 109658522B
- Authority
- CN
- China
- Prior art keywords
- label
- linked list
- node
- voxel
- representative
- Prior art date
- Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
- Active
Links
Images
Classifications
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
- G06T19/00—Manipulating 3D models or images for computer graphics
- G06T19/20—Editing of 3D images, e.g. changing shapes or colours, aligning objects or positioning parts
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
- G06T7/00—Image analysis
- G06T7/10—Segmentation; Edge detection
- G06T7/187—Segmentation; Edge detection involving region growing; involving region merging; involving connected component labelling
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
- G06T2207/00—Indexing scheme for image analysis or image enhancement
- G06T2207/10—Image acquisition modality
- G06T2207/10004—Still image; Photographic image
- G06T2207/10008—Still image; Photographic image from scanner, fax or copier
Landscapes
- Engineering & Computer Science (AREA)
- Physics & Mathematics (AREA)
- Theoretical Computer Science (AREA)
- General Physics & Mathematics (AREA)
- Computer Hardware Design (AREA)
- Software Systems (AREA)
- General Engineering & Computer Science (AREA)
- Computer Graphics (AREA)
- Architecture (AREA)
- Computer Vision & Pattern Recognition (AREA)
- Image Generation (AREA)
- Image Analysis (AREA)
- Apparatus For Radiation Diagnosis (AREA)
Abstract
The invention discloses a simplified method for merging temporary labels for marking a three-dimensional binary image connected body. The method uses a single chain table to store the pixel addresses belonging to the same connected component, and uses a sequence table to store the representative label of each connected component, and the first pixel node and the last pixel node in the single chain table. When the equivalent labels are merged, the merging operation of a plurality of temporary labels can be realized only by merging the single linked list and modifying the corresponding temporary transformed representative labels. The whole merging process has few operation steps, is simple and easy to realize, simplifies the complexity of merging processing of the connected body, reduces the times of scanning the image from the traditional two-time scanning to one-time scanning, improves the overall performance of image processing, and is particularly suitable for real-time processing of large-data-volume images such as three-dimensional images.
Description
Technical Field
The invention relates to the technical field of basic operation research of image processing, in particular to a simplifying method for temporary label merging processing of three-dimensional binary image connected body marks.
Background
In order to identify an object in an image, an object portion of interest in the image is usually distinguished from other portions, that is, a conventional image binarization process is performed. The meaningful region pixels are referred to as object pixels (or foreground pixels, in a three-dimensional binary image, pixels are referred to as voxels), while the other pixels are referred to as background pixels. Since pixels of the same object usually have connectivity, different objects in the image are generally separated by a connected object mark, and then the features of each object are extracted, and then high-level pattern recognition is performed. Therefore, the connected object labeling processing is a precondition for extracting the features of each object in an image, and is one of the most important basic processes in the field of image pattern recognition.
In recent decades, the research on labeling of two-dimensional binary images has been greatly advanced, but the research on labeling of three-dimensional binary images has been almost stopped, mainly because three-dimensional images have not been widely used in the last years. In recent years, however, with the development of three-dimensional imaging technologies such as Computed Tomography (CT), magnetic Resonance Imaging (MRI), optics, ultrasound, sonar, and laser, three-dimensional images have come to be widely used in medical treatment, mineral exploration, 3D printing, and environmental monitoring. In these applications, for example, it is desirable to detect the volume of cancerous tissue, detect the number and volume of blood clots in blood vessels, the number and size of suspended matter in air or water, the number and volume of crystals in material experiments, determine the materials required for a 3D printer by calculating the volume of the components, etc. Therefore, algorithm research of three-dimensional binary image connected body marking processing can provide more powerful technical support for related fields, and with the coming of a big data era, the demand for analysis and identification of mass image data in each field is increasing, so that high-speed research of three-dimensional binary image connected body marking processing is imperative, and the research of the project plays an important role in promoting deep development of the related fields.
Disclosure of Invention
In order to overcome the defects of the prior art, the invention aims to provide a method for simplifying the temporary label merging processing of the three-dimensional binary image connected body label, which makes full use of the random access characteristic of the sequence list and the advantage that the time complexity of single linked list insertion and merging operation is a constant level, simplifies the equivalent merging processing process and saves time for higher-level processing of the three-dimensional binary image.
In order to achieve the purpose, the invention adopts the technical scheme that:
a simplified method for temporary label merging processing of three-dimensional binary image connected body marks comprises the following steps;
(1) Reading a three-dimensional binary image, and storing each voxel value in a one-dimensional array according to coordinate values;
(2) Defining a composite data type DS formed by combining a sequence table and a single-chain table according to the size of the three-dimensional binary image, wherein the sequence table stores a Label value and a pointer field pointing to the single-chain table, the single-chain table connects all object voxels belonging to the same connected body together, the sequence table consists of a data field Label and two pointer fields First and Last, the data field Label stores the Label of the connected body, the pointer field First points to a First node in the single-chain table, and the other pointer field Last points to a Last node in the single-chain table; the data field data of the node forming the single linked list is the coordinate value of the object voxel, and the next pointer field points to the next object voxel belonging to the same connected body;
(3) Defining an array Rep with the storage capacity equal to one fourth of the size of the one-dimensional array for storing the images, and storing the representative labels corresponding to the temporary labels;
(4) Scanning voxels of a one-dimensional array storing a three-dimensional binary image from top to bottom and from left to right, checking whether an object voxel exists in a 26-adjacent voxel working window when the object voxel is scanned, and if not, entering the step (5), otherwise, entering the step (6);
(5) Marking the current object voxel by using a new Label value, respectively storing the Label value and the coordinate value of the current voxel into a Label field of the sequence table of the data type DS established in the step (1) and a data field of a single linked list, setting the value of next of a pointer field of the single linked list to be null, pointing the First pointer field and the Last pointer field of a chain head to nodes of the newly established single linked list, filling the Label value into a representative Label array Rep with a Label value of Label, and returning to the step (4) to continue the marking processing of the subsequent voxel;
(6) And checking each representative label by taking the label of the adjacent voxel as the subscript of the array Rep, selecting the minimum representative label value i from the representative labels, marking the current object voxel by using the label i, generating a new singly linked list node, taking the data field data of the node as the coordinate value of the current voxel, and inserting the node as the last node into a singly linked list pointed by the sequence table element pointer with the label value equal to the label. If the working window has two or more than two representative label values, the temporary label is merged by using the step (7), otherwise, the step (4) is returned to carry out the marking processing of the subsequent voxel;
(7) Determining a label i as a representative label of a current connector label, modifying a pointer field next of a Last node of a current single linked list to point to a first node of a single linked list in a data type DS in which a next equivalent label j is positioned as a first node of a combined single linked list, then enabling a Last node pointer of the single linked list of j to point to a first node of a single linked list in the data type DS in which another equivalent label g participating in combination is positioned, and so on until all the single linked lists of the equivalent labels are combined in one single linked list, and finally modifying a Last pointer in a data type sequence table in which the representative label i is positioned to point to the Last node of the combined single linked list; modifying the First pointer and the Last pointer of the price number nodes such as j, g and the like except for the representative label i to be null, modifying the array element values which take the equivalent labels such as j, g and the like as subscripts in the representative label array Rep by using the representative label i, and returning to the step (4) to carry out marking processing on the subsequent voxels;
(8) Repeating the steps, when all voxels in a pair of three-dimensional binary images are scanned, merging equivalent labels, and finishing the processing, wherein the representative label values of connected bodies in which all object voxels are located and the coordinate values of the voxels are stored in a data type DS formed by combining a sequence table and a single-chain table;
(9) And sequentially reading the coordinate values of the object voxels in the singly linked list, and modifying the label values of the object voxels corresponding to the coordinates by using the representative label values in the sequential list.
The invention has the beneficial effects that:
the three-dimensional binary image connected body marking processing based on raster scanning needs to finish the marking processing of the current object voxel by checking the label condition of the adjacent voxel, and when a working window where the current object voxel is located has a plurality of connected bodies with different labels, the processing of combining equivalent labels needs to be carried out. When object voxels are marked, merging of equivalent labels is needed at any time, and the execution efficiency of the whole marking algorithm is directly influenced by the frequent merging efficiency of equivalent labels. The invention provides a data structure type for storing the representative labels of all connected bodies and the coordinate values of the object voxels in the connected bodies, reduces the merging times of the equivalent labels of the object voxels in the merging process by using the advantages of the inherent operation of the sequence table and the singly linked list data type, and reduces the image scanning times from the traditional two times to 1 time. The efficiency of marking the three-dimensional binary image connected body is improved. Theoretical support and technical reference are provided for the fields of medical image processing, mineral detection, 3D printing and the like in the connected body marking processing.
Drawings
Figure 1-picture of adjacent working window.
FIG. 2 is a flow chart illustrating an equivalent label merging process according to the present invention.
Fig. 3 is a working window diagram of the distribution of foreground voxels v and background voxels.
FIG. 4 is an exemplary diagram of a data structure before the present invention's equivalence label merging process.
FIG. 5 is an exemplary diagram of a data structure after the equivalent label merge process of the present invention.
Detailed Description
The present invention will be described in further detail with reference to the accompanying drawings.
A simplified method for temporary label merging processing of three-dimensional binary image link labels comprises the following steps:
(1) Reading a three-dimensional binary image, and storing each voxel value in a one-dimensional array according to coordinate values;
(2) Defining a composite data variable for equivalent label merging processing, wherein a sequence table stores label values and pointer fields pointing to a singly linked list, and the singly linked list connects all object voxels belonging to the same connected body. The sequence list is composed of a data field R _ Label and two pointer fields First and Last, wherein the data field R _ Label stores the representative labels of the communicating bodies, the pointer field First points to the First node in the single chain list, and the other pointer field Last points to the Last node in the single chain list; the nodes that make up the singly linked list contain object voxel coordinate values (XSize, YSize), and the pointer field next points to the next object voxel that belongs to the same connected volume.
(3) And defining an array Rep with the storage capacity equal to one fourth of the size of the one-dimensional array for storing the images, wherein the array Rep is used for storing the representative labels corresponding to the temporary labels.
(4) Scanning voxels of a one-dimensional array storing a three-dimensional binary image from top to bottom and from left to right, checking whether the adjacent voxels have the object voxels in a 26-adjacent voxel working window shown in figure 1 when one object voxel is scanned, and if not, entering the step (5), otherwise, entering the step (6);
(5) Marking the current object voxel by using a new Label value l, respectively storing the Label value and the coordinate value of the current voxel into the R _ Label field of the sequence table established in the step (2) and the data field of the single linked list, setting the value of the pointer field next of the single linked list to be null, pointing the pointers First and Last to the node of the newly established single linked list, and filling the value l into the representative Label array Rep with the lower Label value l. Returning to the step (4) to continue the marking processing of the subsequent voxels;
(6) The label of the adjacent voxel is used as the index of the array Rep to check each representative label, and the minimum representative label value i is selected from the labels, the label i is used for marking the current object voxel, a new singly linked list node is generated, the coordinate value of the current voxel is stored in the data field of the node, and the node is used as the last node to be inserted into the singly linked list pointed by the sequence table element pointer with the label value equal to the label. And (7) if two or more representative label values exist in the working window, carrying out temporary label combination processing by using the step (7). Otherwise, returning to the step (4) to mark the subsequent voxels;
(7) The reference numeral i is determined as a representative reference numeral of the current via number. i is the first node in the single linked list corresponding to the data type of the I, the pointer field next of the Last node of the current single linked list is modified to point to the first node of the single linked list of the next equivalent label j in the data type, then the Last node pointer of the single linked list of j points to the first node of the single linked list of the other data type of the equivalent label g participating in the combination, and so on until all the single linked lists of the equivalent labels are combined in one single linked list, and finally the Last pointer in the data type sequence list representing the label i points to the Last node of the combined single linked list is modified; modifying the First pointer and the Last pointer of the price mark nodes except for the mark i, such as j, g and the like to be null, and modifying the array element values which represent the equivalent marks of the mark i, such as j, g and the like in the mark array Rep by using the mark i. And (5) returning to the step (4) to carry out marking processing of the subsequent voxel.
(8) Repeating the steps, when all voxels in a pair of three-dimensional binary images are scanned, merging equivalent labels, and finishing the processing, wherein the representative label values of connected bodies in which all object voxels are located and the coordinate values of the voxels are stored in a data type formed by combining a sequence table and a single chain table;
(9) And reading the coordinate values of the object voxels in the singly linked list in sequence, and modifying the label values of the object voxels corresponding to the coordinates by using the representative label values in the sequence list.
Example (b):
reading a three-dimensional binary image, and sequentially storing each voxel into a one-dimensional array. According to fig. 1, data variables for storing label values and statistical connected body voxels required for merging equivalent labels are defined, and an array with a storage capacity of one quarter of the size of a three-dimensional binary image is further defined for storing a representative label of each label. Scanning the image line by line from top to bottom, from left to right, when an object voxel is scanned, checking whether the object voxel exists in the adjacent voxel according to a 26-adjacent working window shown in fig. 1, if not, entering the step (1), otherwise, entering the step (2), wherein a specific processing schematic diagram is shown in fig. 2;
(1) Marking the current object voxel by using a new Label value, respectively storing the Label value l and the coordinate value of the current voxel into an R _ Label field of a sequence table of a combined data type and XSize and YSize of a single chain table, setting the value of the next of a pointer field of the single chain table to be null, pointing the First pointer field and the Last pointer field of the chain head to the node of the newly established single chain table, and filling the value l into a representative Label array Rep with the index value l. Continuing the marking treatment of the subsequent voxels;
(2) The label of adjacent voxel is used as index of array Rep to check each representative label, and select the minimum representative label value, and use the label value to mark the current object voxel, and generate a new single-linked list node, store the coordinate value of current voxel, and insert the node as the last node into the single-linked list pointed by the sequence list element pointer whose label value is equal to the label. And (4) if two or more representative label values exist in the working window, carrying out the combination processing of the temporary labels by using the step (3).
(3) The reference numeral i is determined as a representative reference numeral of the current via number. i is the first node in the single linked list corresponding to the data type where i is located, the pointer field next of the Last node of the current single linked list is modified to point to the first node of the single linked list of the data type where the next equivalent label j is located, then the Last node pointer of the single linked list of j is made to point to the first node of the single linked list of the data type where the other equivalent label g participating in the combination is located, and so on until all the single linked lists of the equivalent labels are combined in one single linked list, and finally the Last pointer in the data type sequence table representing the label i is modified to point to the Last node of the single linked list after the combination; modifying the First pointer and the Last pointer of the price number nodes such as j, g and the like except for the index i to be null, and modifying the array element values taking the equivalent indexes such as j, g and the like as subscripts in the index array Rep by using the index i.
In the working window shown in fig. 3, the current voxel is an object voxel, and the object voxels adjacent to the current voxel have v 4 ，v 5 ，v 7 ，v 11 The label values and the singly linked list data values of the data structure table are shown in fig. 4 (a) for four adjacent voxels before labeling the current object voxel, since the object voxel v 4 And v 7 Adjacent to each other, so that there are three equivalent labels whose representative labels are 7,2 and 10, respectively, as shown in fig. 4 (b), when marking the current object voxel, a node is generated, the coordinate values (XSize, YSize) of the current object voxel are stored in the data field of the node, and the smallest representative label is selected from the three equivalent labelsNumber 2, linking the node into the single chain table with label value 2 by tail insertion, making its next pointer point to the First node in the single chain table with label 7 to be equivalently combined, then modifying the next pointer of the Last node of the single chain table with label 7 to point to the First node in the single chain table with label 10, finally modifying the Last pointer of label 2 to point to the Last node in the single chain table with label 10, so merging all voxels in the equivalent labels to be combined into one single chain table, finally modifying the First and Last pointers of the head nodes with labels 7 and 10 to null, and combining the data values with equivalent labels as shown in fig. 5 (a). The values of the indices 7 and 10 in the array of representative indices are modified with the minimum representative index value of 2, respectively, as shown in fig. 5 (b). Thus, the process of marking the voxel of an object and combining equivalent labels is completed.
And after one-time scanning is finished, sequentially reading R _ Label data in the sequence table in the generated data structure table, and assigning the Label of the object voxel of each node in the corresponding single linked list. Because the singly linked list stores the addresses of all object voxels in the image, the invention can complete the marking of the whole image by only scanning the image once. Scanning an image is a very time-consuming operation, and it is self-evident that reducing the number of operations to scan an image once (especially for real-time images, images with large pixel values) increases the efficiency of the overall image processing.
The merged equivalent label processing of the present invention is an indispensable operation in object voxel labeling processing and is a key operation that affects the efficiency of the connected body labeling processing. In order to meet the actual requirement of high speed of three-dimensional binary image processing, in the object voxel marking process, the invention provides a new data model formed by combining a sequence table and a single chain table. When the merging processing of the equivalent labels is needed, only the merging operation is needed to be carried out on the single linked list, and the merging processing steps are few, simple and easy to realize. The invention fully utilizes the random access characteristic of the sequence list and the advantage that the time complexity of the single linked list insertion and combination operation is constant level, simplifies the equivalent combination processing process and saves time for the higher-level processing of the three-dimensional binary image.
Claims (1)
1. A simplified method for temporary label merging processing of three-dimensional binary image connected body labels is characterized by comprising the following steps;
(1) Reading a three-dimensional binary image, and storing each voxel value in a one-dimensional array according to coordinate values;
(2) Defining a composite data type DS formed by combining a sequence table and a single-chain table according to the size of a three-dimensional binary image, wherein the sequence table stores a Label value and a pointer field pointing to the single-chain table, the single-chain table connects all object voxels belonging to the same connected body together, the sequence table is composed of a data field Label and two pointer fields First and Last, the data field Label stores the Label of the connected body, the pointer field First points to a First node in the single-chain table, and the other pointer field Last points to a Last node in the single-chain table; the data field data of the node forming the single linked list is an object voxel coordinate value, and the pointer field next points to the next object voxel belonging to the same connected body;
(3) Defining an array Rep with the storage capacity equal to one fourth of the size of the one-dimensional array for storing the images, and storing the representative labels corresponding to the temporary labels;
(4) Scanning voxels of the one-dimensional array storing the three-dimensional binary image from top to bottom and from left to right, checking whether an object voxel exists in a 26-adjacent voxel working window when the object voxel is scanned, and if not, entering the step (5), otherwise, entering the step (6);
(5) Marking the current object voxel by using a new Label value, respectively storing the Label value and the coordinate value of the current voxel into a Label field of the sequence table of the data type DS established in the step (1) and a data field of the single linked list, setting the next value of a pointer field of the single linked list to be null, pointing the First pointer field and the Last pointer field of the chain head to the node of the newly established single linked list, filling the Label value into a representative Label array Rep with the Label value of Label, and returning to the step (4) to continue the marking processing of the subsequent voxel;
(6) Checking each representative label by taking the label of an adjacent voxel as a subscript of an array Rep, selecting the minimum representative label value i from the representative labels, marking the current object voxel by the label i, generating a new singly linked list node, taking the data field data of the node as the coordinate value of the current voxel, inserting the node as the last node into a singly linked list with the label value equal to the pointed direction of the sequence table element pointer of the label, if two or more representative label values exist in a working window, merging the temporary labels by using the step (7), otherwise, returning to the step (4) to mark the subsequent voxel;
(7) Determining a label i as a representative label of a current connector label, modifying a pointer field next of a Last node of a current single linked list to point to a first node of a single linked list in a data type DS in which a next equivalent label j is positioned as a first node of a combined single linked list, then enabling a Last node pointer of the single linked list of j to point to a first node of a single linked list in the data type DS in which another equivalent label g participating in combination is positioned, and so on until all the single linked lists of the equivalent labels are combined in one single linked list, and finally modifying a Last pointer in a data type sequence table in which the representative label i is positioned to point to the Last node of the combined single linked list; modifying the First pointer and the Last pointer of the price number nodes such as j, g and the like except for the representative label i to be null, modifying the array element values which take the equivalent labels such as j, g and the like as subscripts in the representative label array Rep by using the representative label i, and returning to the step (4) to carry out marking processing on the subsequent voxels;
(8) Repeating the steps, and after all voxels in a pair of three-dimensional binary images are scanned, merging equivalent labels, ending the processing, and storing the representative label values of the connected bodies in which all object voxels are located and the coordinate values of the voxels in the images into a data type DS formed by combining a sequence table and a single-chain table;
(9) And sequentially reading the coordinate values of the object voxels in the singly linked list, and modifying the label values of the object voxels corresponding to the coordinates by using the representative label values in the sequential list.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201811427359.1A CN109658522B (en) | 2018-11-27 | 2018-11-27 | Simplified method for temporary label merging processing of three-dimensional binary image connected body label |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201811427359.1A CN109658522B (en) | 2018-11-27 | 2018-11-27 | Simplified method for temporary label merging processing of three-dimensional binary image connected body label |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN109658522A CN109658522A (en) | 2019-04-19 |
| CN109658522B true CN109658522B (en) | 2023-03-14 |
Family
ID=66111553
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN201811427359.1A Active CN109658522B (en) | 2018-11-27 | 2018-11-27 | Simplified method for temporary label merging processing of three-dimensional binary image connected body label |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN109658522B (en) |
Families Citing this family (1)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN115237369B (en) * | 2022-09-23 | 2022-12-13 | 成都博宇利华科技有限公司 | High-precision information stamp marking method |
Citations (2)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN102402796A (en) * | 2011-10-26 | 2012-04-04 | 重庆大学 | Three-dimensional structural description method of hepatic vascular system |
| CN108257118A (en) * | 2018-01-08 | 2018-07-06 | 浙江大学 | A kind of fracture adhesion dividing method based on normal direction corrosion and random walk |
Family Cites Families (1)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| US7697759B2 (en) * | 2004-05-11 | 2010-04-13 | The United States Of America As Represented By The Administrator Of The National Aeronautics And Space Administration | Split-remerge method for eliminating processing window artifacts in recursive hierarchical segmentation |
-
2018
- 2018-11-27 CN CN201811427359.1A patent/CN109658522B/en active Active
Patent Citations (2)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN102402796A (en) * | 2011-10-26 | 2012-04-04 | 重庆大学 | Three-dimensional structural description method of hepatic vascular system |
| CN108257118A (en) * | 2018-01-08 | 2018-07-06 | 浙江大学 | A kind of fracture adhesion dividing method based on normal direction corrosion and random walk |
Non-Patent Citations (2)
| Title |
|---|
| 基于3D激光雷达的无人水面艇海上目标检测;李小毛等;《上海大学学报(自然科学版)》;20170228(第01期);全文 * |
| 基于标签化医学图像的多阈值三维重建算法;肖洪旭等;《计算机工程与科学》;20171015(第10期);全文 * |
Also Published As
| Publication number | Publication date |
|---|---|
| CN109658522A (en) | 2019-04-19 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| US11710242B2 (en) | Methods and systems for image segmentation | |
| Zhang et al. | Mesh segmentation via recursive and visually salient spectral cuts | |
| Yan et al. | Variational mesh segmentation via quadric surface fitting | |
| CN108921939A (en) | A kind of method for reconstructing three-dimensional scene based on picture | |
| BRPI0606705B1 (en) | METHOD FOR IDENTIFYING OBJECTS IN AN IMAGE | |
| CN1136848A (en) | Fingerprint matching system | |
| CN109191510B (en) | 3D reconstruction method and device for pathological section | |
| JP4248399B2 (en) | Automatic branch labeling method | |
| CN108010123A (en) | A kind of three-dimensional point cloud acquisition methods for retaining topology information | |
| CN110009745B (en) | Method for extracting plane from point cloud according to plane element and model drive | |
| CN100375124C (en) | A skeletonized object rebuild method | |
| CN108716916A (en) | A kind of distributed parallel asterism barycenter extracting method and FPGA realization devices based on superblock | |
| Hua et al. | Geodesic distance-weighted shape vector image diffusion | |
| Gao et al. | Classification of 3D terracotta warrior fragments based on deep learning and template guidance | |
| CN109658522B (en) | Simplified method for temporary label merging processing of three-dimensional binary image connected body label | |
| Xu et al. | Accurate polygonal mapping of buildings in satellite imagery | |
| CN106023094B (en) | Bone tissue microstructure repair system based on image and its restorative procedure | |
| Wang et al. | Multispectral point cloud superpoint segmentation | |
| CN110751732B (en) | Method for converting 2D image into 3D image | |
| CN109872353B (en) | White light data and CT data registration method based on improved iterative closest point algorithm | |
| CN114494637B (en) | Sandstone three-dimensional real model reconstruction method based on structural matrix | |
| Herbert et al. | Contour correspondence for serial section reconstruction: complex scenarios in palaeontology | |
| CN113516669B (en) | CT image-based trachea extraction method, device, equipment and storage medium | |
| He et al. | 3D reconstruction of Chinese hickory trees for mechanical harvest | |
| CN117830261A (en) | Physiological channel labeling method, system, device and nonvolatile storage medium |
Legal Events
| Date | Code | Title | Description |
|---|---|---|---|
| PB01 | Publication | ||
| PB01 | Publication | ||
| SE01 | Entry into force of request for substantive examination | ||
| SE01 | Entry into force of request for substantive examination | ||
| GR01 | Patent grant | ||
| GR01 | Patent grant |