CN109299665B - Human-shaped contour description method based on LSD algorithm - Google Patents
Human-shaped contour description method based on LSD algorithm Download PDFInfo
- Publication number
- CN109299665B CN109299665B CN201810998054.XA CN201810998054A CN109299665B CN 109299665 B CN109299665 B CN 109299665B CN 201810998054 A CN201810998054 A CN 201810998054A CN 109299665 B CN109299665 B CN 109299665B
- Authority
- CN
- China
- Prior art keywords
- main direction
- point
- executing
- grid
- gradient
- 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
Classifications
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06V—IMAGE OR VIDEO RECOGNITION OR UNDERSTANDING
- G06V40/00—Recognition of biometric, human-related or animal-related patterns in image or video data
- G06V40/20—Movements or behaviour, e.g. gesture recognition
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06V—IMAGE OR VIDEO RECOGNITION OR UNDERSTANDING
- G06V10/00—Arrangements for image or video recognition or understanding
- G06V10/40—Extraction of image or video features
- G06V10/44—Local feature extraction by analysis of parts of the pattern, e.g. by detecting edges, contours, loops, corners, strokes or intersections; Connectivity analysis, e.g. of connected components
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06V—IMAGE OR VIDEO RECOGNITION OR UNDERSTANDING
- G06V40/00—Recognition of biometric, human-related or animal-related patterns in image or video data
- G06V40/10—Human or animal bodies, e.g. vehicle occupants or pedestrians; Body parts, e.g. hands
Landscapes
- Engineering & Computer Science (AREA)
- Physics & Mathematics (AREA)
- General Physics & Mathematics (AREA)
- Multimedia (AREA)
- Theoretical Computer Science (AREA)
- Human Computer Interaction (AREA)
- Computer Vision & Pattern Recognition (AREA)
- Health & Medical Sciences (AREA)
- General Health & Medical Sciences (AREA)
- Psychiatry (AREA)
- Social Psychology (AREA)
- Image Analysis (AREA)
Abstract
The invention relates to a human-shaped contour description method based on an LSD algorithm, which is characterized by comprising the following steps of: inputting data, calculating gradient information of the moving points, gridding, traversing the moving points, updating main direction points of the grids, traversing the moving points again, updating the main direction of the grids by using an LSD algorithm, calculating line segment information in all the grids and outputting data; the method has the advantages of low time consumption, capability of judging the human shape of each visual angle, and suitability for the characteristics of various visual angles for monitoring and shooting in stores, low contrast in an infrared mode and complex scene.
Description
Technical Field
The method designs the field of image processing, and belongs to an image feature extraction method in the field of image analysis.
Background
At present, the method is the most main human figure feature extraction method based on the HOG feature and the CNN deep convolution feature, the two feature description methods are high in dimensionality, and human figures need to be judged by combining a machine learning algorithm, so that on one hand, the time consumption of target detection is high, and the requirements of real-time application cannot be met, and on the other hand, a judgment model obtained by the machine learning algorithm cannot be suitable for security multi-view application scenes; the actual shop monitoring has the characteristics of complex scene, various shooting visual angles and low image quality in an infrared mode, and the original HOG characteristic and CNN deep convolution characteristic cannot meet the application requirements of the actual shop monitoring.
Disclosure of Invention
In view of this, the present invention provides a human-shaped outline description method based on an LSD algorithm, which is characterized by comprising the following steps:
(1) Inputting data, wherein the data is a motion point set detected by a motion target, and executing the step (2);
(2) Calculating gradient information of the moving point, calculating gradient value and gradient direction of the moving point by using the gray-scale map, normalizing the gradient direction to the range of [0,180], and executing the step (3);
(3) Performing gridding treatment, namely performing spatial 8*8 gridding segmentation on the motion point set, initializing a main direction point in a grid to be null, and executing the step (4), wherein the main direction is-1;
(4) Traversing the motion points, if the traversal is finished, executing the step (6), otherwise, acquiring the current motion point P and executing the step (5);
(5) Updating the main direction point of the grid, calculating the gridding coordinate (Yp, xp) of the current moving point P, if the main direction in the grid (Yp, xp) is-1 or the gradient value Grad (P0) of the existing main direction point P0 is smaller than the gradient value Grad (P) of the current moving point P, updating the main direction in the grid (Yp, xp) to be the gradient direction of the moving point P, updating the main direction point of the grid to be P, and continuing to execute the step (4);
(6) Traversing the motion points again, if the traversing is finished, executing the step (8), otherwise, acquiring the current motion point P, executing the step (7), and executing the step (8);
(7) Updating the main direction of the grid by using an LSD algorithm, calculating gridding coordinates (Yp, xp) of the current moving point P, if the difference between the gradient direction of the main direction in the grid (Yp, xp) and the gradient direction of the current moving point P is less than 30 degrees, updating the main direction in the grid (Yp, xp) by using a direction merging method in the LSD algorithm, accumulating the current moving point P into the main direction point of the grid (Yp, xp), and continuously executing the step (6);
(8) Calculating the line segment information in all the grids, initializing a human-shaped outline information linked list L to be empty, traversing all the grids, calculating the local line segment information in the grids, wherein the local line segment information comprises the direction, gradient size and central position information of the line segments in the grids, if the number of main direction points in the grids is more than 3, considering that the local line segments in the grids are effective, inserting the effective local line segment information into the outline information linked list L, and executing the step (9) after traversing is finished;
(9) And outputting data and outputting a human-shaped profile information linked list L consisting of local lines.
The beneficial results of the invention are as follows: according to the method, the human-shaped outline characteristics are extracted through the LSD algorithm on the basis of the motion points, on one hand, the interference of redundant motion points on the human-shaped outline is reduced, the algorithm robustness is provided, on the other hand, the human-shaped outline is described in a local line segment mode, local line segment information comprises local direction and strength information, human shape judgment is conveniently carried out in a mode judgment mode, the human shape judgment efficiency is high, the method is suitable for monitoring application scenes of multiple visual angles in a market, and the method has a wide market prospect and an application value.
Detailed Description
In order to make the technical problems, technical solutions and advantageous effects to be solved by the present invention more apparent, the present invention is described in detail below with reference to embodiments. It should be noted that the specific embodiments described herein are only for illustrating the present invention and are not to be construed as limiting the present invention, and the equivalent substitutions and modifications of the products that can achieve the same functions are included in the scope of the present invention. The specific method comprises the following steps:
example 1: in view of this, the present invention provides a human-shaped outline description method based on an LSD algorithm, which is characterized by comprising the following steps:
(1) Inputting data, wherein the data is a motion point set detected by a motion target, and executing the step (2);
(2) Calculating gradient information of the moving point, calculating gradient value and gradient direction of the moving point by using the gray-scale map, normalizing the gradient direction to the range of [0,180], and executing the step (3);
(3) Performing gridding treatment, namely performing spatial 8*8 gridding segmentation on the motion point set, initializing a main direction point in a grid to be null, and executing the step (4), wherein the main direction is-1;
(4) Traversing the motion points, if the traversal is finished, executing the step (6), otherwise, acquiring the current motion point P and executing the step (5);
(5) Updating the main direction point of the grid, calculating the gridding coordinate (Yp, xp) of the current moving point P, if the main direction in the grid (Yp, xp) is-1 or the gradient value Grad (P0) of the existing main direction point P0 is smaller than the gradient value Grad (P) of the current moving point P, updating the main direction in the grid (Yp, xp) to be the gradient direction of the moving point P, updating the main direction point of the grid to be P, and continuing to execute the step (4);
(6) Traversing the motion points again, if the traversal is finished, executing the step (8), otherwise acquiring the current motion point P, executing the step (7), and executing the step (8);
(7) Updating the main direction of the grid by using an LSD algorithm, calculating gridding coordinates (Yp, xp) of the current moving point P, if the difference between the gradient direction of the main direction in the grid (Yp, xp) and the gradient direction of the current moving point P is less than 30 degrees, updating the main direction in the grid (Yp, xp) by using a direction combination method in the LSD algorithm, accumulating the current moving point P into the main direction point of the grid (Yp, xp), and continuously executing the step (6);
(8) Calculating the line segment information in all the grids, initializing a human-shaped outline information linked list L to be empty, traversing all the grids, calculating the local line segment information in the grids, wherein the local line segment information comprises the direction, gradient size and central position information of the line segments in the grids, if the number of main direction points in the grids is more than 3, considering that the local line segments in the grids are effective, inserting the effective local line segment information into the outline information linked list L, and executing the step (9) after the traversal is finished;
(9) And outputting data and outputting a human-shaped profile information linked list L consisting of local lines.
Claims (1)
1. The invention relates to a human-shaped contour description method based on an LSD algorithm, which is characterized by comprising the following steps of:
(1) Inputting data which is a set of motion points detected by a motion target, and executing the step (2);
(2) Calculating gradient information of the moving points, calculating gradient values and gradient directions of the moving points by using a gray scale map, and normalizing the gradient directions to a range of [0,180], and executing the step (3);
(3) Performing gridding processing, namely performing spatial 8*8 gridding segmentation on the motion point set, initializing a main direction as-1 and setting a main direction as null in the grid, and executing the step (4);
(4) Traversing the motion points, if the traversal is finished, executing the step (6), otherwise, acquiring the current motion point P, and executing the step (5);
(5) Updating the main direction point of the grid, calculating the gridding coordinate (Yp, xp) of the current moving point P, if the main direction in the grid (Yp, xp) is-1 or the gradient value Grad (P0) of the existing main direction point P0 is smaller than the gradient value Grad (P) of the current moving point P, updating the main direction in the grid (Yp, xp) to be the gradient direction of the moving point P, updating the main direction point of the grid to be P, and continuing to execute the step (4);
(6) Traversing the motion points again, if the traversal is finished, executing the step (8), otherwise, acquiring the current motion point P, executing the step (7), and executing the step (8);
(7) Updating the main direction of the grid by using an LSD algorithm, calculating the gridding coordinate (Yp, xp) of the current motion point P, if the difference between the gradient direction of the main direction in the grid (Yp, xp) and the gradient direction of the current motion point P is less than 30 degrees, updating the main direction in the grid (Yp, xp) by using a direction combination method in the LSD algorithm, accumulating the current motion point P into the main direction point of the grid (Yp, xp), and continuously executing the step (6);
(8) Calculating the line segment information in all the grids, initializing a humanoid outline information linked list L to be empty, traversing all the grids, calculating the local line segment information in the grids, including the direction, gradient size and central position information of the line segments in the grids, if the number of main direction points in the grids is more than 3, considering that the local line segments in the grids are effective, inserting the effective local line segment information into the outline information linked list L, and executing the step (9) after traversing is finished;
(9) And outputting data, namely outputting the human-shaped profile information linked list L consisting of the local lines.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201810998054.XA CN109299665B (en) | 2018-08-29 | 2018-08-29 | Human-shaped contour description method based on LSD algorithm |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201810998054.XA CN109299665B (en) | 2018-08-29 | 2018-08-29 | Human-shaped contour description method based on LSD algorithm |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN109299665A CN109299665A (en) | 2019-02-01 |
| CN109299665B true CN109299665B (en) | 2023-04-14 |
Family
ID=65165954
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN201810998054.XA Active CN109299665B (en) | 2018-08-29 | 2018-08-29 | Human-shaped contour description method based on LSD algorithm |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN109299665B (en) |
Citations (3)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| JPH10143654A (en) * | 1996-11-08 | 1998-05-29 | Sony Corp | Outline extracting device and outline extracting method |
| CN105550678A (en) * | 2016-02-03 | 2016-05-04 | 武汉大学 | Human body motion feature extraction method based on global remarkable edge area |
| CN107194940A (en) * | 2017-05-23 | 2017-09-22 | 北京计算机技术及应用研究所 | A kind of coloured image contour extraction method based on color space and line segment |
-
2018
- 2018-08-29 CN CN201810998054.XA patent/CN109299665B/en active Active
Patent Citations (3)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| JPH10143654A (en) * | 1996-11-08 | 1998-05-29 | Sony Corp | Outline extracting device and outline extracting method |
| CN105550678A (en) * | 2016-02-03 | 2016-05-04 | 武汉大学 | Human body motion feature extraction method based on global remarkable edge area |
| CN107194940A (en) * | 2017-05-23 | 2017-09-22 | 北京计算机技术及应用研究所 | A kind of coloured image contour extraction method based on color space and line segment |
Non-Patent Citations (1)
| Title |
|---|
| 多轮廓三维立体视景图像的投影检测算法;睢丹等;《吉林大学学报(理学版)》;20160726(第04期);全文 * |
Also Published As
| Publication number | Publication date |
|---|---|
| CN109299665A (en) | 2019-02-01 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN107093205B (en) | A kind of three-dimensional space building window detection method for reconstructing based on unmanned plane image | |
| CN106709950B (en) | Binocular vision-based inspection robot obstacle crossing wire positioning method | |
| CN104008553B (en) | Crack detection method with image gradient information and watershed method conflated | |
| CN108090572B (en) | Control method of offshore wind farm augmented reality system | |
| CN104732536A (en) | Sub-pixel edge detection method based on improved morphology | |
| CN104299260A (en) | Contact network three-dimensional reconstruction method based on SIFT and LBP point cloud registration | |
| CN108537808A (en) | A kind of gluing online test method based on robot teaching point information | |
| CN104268519B (en) | Image recognition terminal and its recognition methods based on pattern match | |
| CN102568007A (en) | Image processing method performing quasi-circular detection by improving Hough transformation | |
| CN109087323A (en) | A kind of image three-dimensional vehicle Attitude estimation method based on fine CAD model | |
| CN103778411B (en) | Circle detection method and device based on raster image division | |
| CN104933738A (en) | Visual saliency map generation method based on local structure detection and contrast | |
| Pei et al. | Effective image registration based on improved harris corner detection | |
| KR101460313B1 (en) | Apparatus and method for robot localization using visual feature and geometric constraints | |
| CN109544575B (en) | Three-dimensional contour reconstruction method based on ISAR sequence polygon matching | |
| CN108109208B (en) | Augmented reality method for offshore wind farm | |
| CN210386980U (en) | Machine vision-based intelligent cooling bed control system | |
| CN104992448A (en) | Automatic positioning method for robot picking grapes in a damage-free way | |
| CN112231848B (en) | Method and system for constructing vehicle spraying model | |
| CN104408726B (en) | A kind of image outline matching process based on triangle | |
| CN109299665B (en) | Human-shaped contour description method based on LSD algorithm | |
| Ji et al. | DRV-SLAM: An Adaptive Real-Time Semantic Visual SLAM Based on Instance Segmentation Toward Dynamic Environments | |
| Yang et al. | Cherry recognition based on color channel transform | |
| CN110660081B (en) | Target tracking method based on self-adaptive feature selection | |
| CN104766085B (en) | A kind of multiple dimensioned pattern recognition method |
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 | ||
| PE01 | Entry into force of the registration of the contract for pledge of patent right | ||
| PE01 | Entry into force of the registration of the contract for pledge of patent right |
Denomination of invention: A Method for Describing Human Figure Contours Based on LSD Algorithm Effective date of registration: 20230701 Granted publication date: 20230414 Pledgee: Bank of Beijing Limited by Share Ltd. Shanghai branch Pledgor: SHANGHAI ULUCU ELECTRONIC TECHNOLOGY CO.,LTD. Registration number: Y2023980046838 |