CN112200879B

CN112200879B - Map lightweight compression transmission method

Info

Publication number: CN112200879B
Application number: CN202011193070.5A
Authority: CN
Inventors: 牛成成; 傅建中; 林志伟; 沈洪垚
Original assignee: Zhejiang University ZJU
Current assignee: Zhejiang University ZJU
Priority date: 2020-10-30
Filing date: 2020-10-30
Publication date: 2022-06-14
Anticipated expiration: 2040-10-30
Also published as: CN112200879A

Abstract

The invention relates to a map lightweight compression transmission method, and belongs to the technical field of synchronous positioning and map building SLAM. The method comprises the steps of map compression of a Linux server side and map decompression of a Windows client side, wherein the map compression step of the Linux server side comprises the following steps: 1) acquiring a 2D grid map established at a certain moment, and acquiring the overall size of the map; 2) acquiring the outer contour and the inner contour of the 2D grid map, and extracting black pixels of the 2D grid map; 3) outputting the outer contour, the inner contour, the black pixels and the overall size of the 2D grid map; the map decompression step of the Windows client comprises the following steps: 4) acquiring the outer contour, the inner contour, black pixels and the overall size of the 2D grid map; 5) extracting a blank gray map generated according to the overall size of the 2D grid map; 6) filling white pixels in the outer contour, filling gray pixels in the inner contour, and filling black pixels obtained in the step 4) between the outer contour and the inner contour; 7) and obtaining the decompressed 2D grid map.

Description

Map lightweight compression transmission method

Technical Field

The invention relates to the technical field of synchronous positioning and map building SLAM, in particular to a map lightweight compression transmission method.

Background

With the rapid development of computer technology, sensor technology and network transmission technology, more and more fields of SLAM appear, and the increase of the demand for automation in various industries keeps the SLAM in strong growth.

SLAM technology (Simultaneous localization and mapping) is a Simultaneous localization and mapping technology. SLAM technology can sense the environment and map and locate simultaneously with onboard sensors in partially known or unknown environments.

Two technical difficulties are critical for SLAM: and (3) establishing a map and navigating, wherein the map establishing algorithm utilizes data from sensors such as an IMU (inertial measurement Unit), a speedometer, a laser radar and a camera to establish a map of a certain scene. Navigation is autonomous navigation of a target point based on a known global map or a partially known map.

There are a number of modern SLAM mapping algorithms that require a priori knowledge about each algorithm to select which algorithm to use in a particular problem. The invention patent application publication No. CN110608742A discloses a map construction method and device based on a particle filter SLAM, which judges whether laser sensor data is suitable or not by scanning and matching the laser sensor data through an iterative closest point algorithm; if the scanning matching fails, adopting odometer data, and if the scanning matching succeeds, adopting laser sensor data; sampling and updating a map by combining the data of the laser sensor and the data of the milemeter; and determining whether to perform resampling by judging whether the effective particle number meets the condition.

The mapping algorithm inputs data of sensors such as an IMU (inertial measurement Unit), a speedometer, a laser radar and a camera, outputs a map of a certain scene, and has the requirement that the output map is as close to a real map of the ground as possible.

How to evaluate the difference from the real map is a challenging problem. The higher the resolution of the map, the closer to the real map, which is a good criterion for a map. The higher the map resolution, the higher the accuracy of the map is inherently improved, but the higher the requirements on hardware and network transmission for running the mapping algorithm. Particularly, when map data needs to be transmitted across systems in Linux and Windows, higher requirements are put on a network. A sheet of 100X 100m²The map of (2) which has a resolution of 0.01M and which will contain 1 hundred million pixels is converted into a map of uncompressed PGM cells, the size of which is close to 100M, which is greatly disadvantageous for its transmission in a network, and particularly requires visualization across Linux and Windows systems during the mapping process.

Disclosure of Invention

The invention aims to provide a map lightweight compression transmission method, which solves the problem that a map occupies a large amount of network resources in the Linux and Windows transmission process.

In order to achieve the above object, the map lightweight compression transmission method provided by the present invention includes a map compression step at the Linux server side and a map decompression step at the Windows client side, wherein the map compression step at the Linux server side includes:

1) acquiring a 2D grid map established at a certain moment, and acquiring the overall size of the map;

2) acquiring the outer contour and the inner contour of the 2D grid map, and extracting black pixels of the 2D grid map;

3) outputting the outer contour, the inner contour, the black pixels and the overall size of the 2D grid map;

the map decompression step of the Windows client comprises the following steps:

4) acquiring the outer contour, the inner contour, black pixels and the overall size of the 2D grid map;

5) extracting a blank gray map generated according to the overall size of the 2D grid map;

6) filling white pixels in the outer contour, filling gray pixels in the inner contour, and filling black pixels obtained in the step 4) between the outer contour and the inner contour;

7) and obtaining the decompressed 2D grid map.

In the above technical solution, the outline refers to a boundary between an outermost gray area in the original map and an area other than the map. The inner contour refers to a gray area completely or mostly surrounded by white pixels. The map is compressed at a Linux end, then the compressed map is transmitted to the Windows through a network, and decompression and visualization are performed under the Windows. Under the condition of small visual influence, the network transmission cost of the map is greatly reduced, and a feasible solution is provided for map transmission with high precision and large area.

The method greatly reduces the size of the map in cross-system transmission under the condition of not influencing visualization by compressing the map of a Linux server side and decompressing the map at a Windows client side, provides technical support for visualization of high-precision maps, and improves the timeliness and stability of transmission.

Drawings

FIG. 1 is a flowchart of a map compression method at the Linux server side according to an embodiment of the present invention;

FIG. 2 is a flowchart of a map decompression method of a Windows client in an embodiment of the present invention;

FIG. 3 is a 2D grid map to be compressed of the Linux server according to the embodiment of the present invention;

FIG. 4 is a schematic diagram of a binarized 2D grid map in an embodiment of the present invention;

FIG. 5 is a schematic diagram illustrating an outline extraction result of a 2D grid map according to an embodiment of the present invention;

FIG. 6 is a schematic diagram of a 2D grid map after isolated points are selectively removed according to an embodiment of the present disclosure;

FIG. 7 is a schematic diagram illustrating an inner contour extraction result of a 2D grid map according to an embodiment of the present invention;

FIG. 8 is a schematic diagram of all outlines of a 2D grid map in an embodiment of the present invention;

FIG. 9 is a diagram illustrating the reconstruction effect of the test map 1 according to an embodiment of the present invention;

FIG. 10(a) is a schematic diagram of a test map 2 according to an embodiment of the present invention, and (b) is a reconstruction effect diagram thereof;

fig. 11(a) is a schematic diagram of the test map 3 in the embodiment of the present invention, and (b) is a reconstruction effect diagram thereof.

Detailed Description

In order to make the objects, technical solutions and advantages of the present invention more apparent, the present invention will be further described with reference to the following embodiments and accompanying drawings. It is to be understood that the embodiments described are only a few embodiments of the present invention, and not all embodiments. All other embodiments, which can be derived by a person skilled in the art from the described embodiments without any inventive step, are within the scope of protection of the invention.

Unless defined otherwise, technical or scientific terms used herein shall have the ordinary meaning as understood by one of ordinary skill in the art to which this invention belongs. The use of "including" or "comprising" and the like in the present invention is intended to mean that the elements or items listed before the word "comprise" or "comprising" and the like, include the elements or items listed after the word and their equivalents, but do not exclude other elements or items. The terms "connected" or "coupled" and the like are not restricted to physical or mechanical connections, but may include electrical connections, whether direct or indirect. "upper", "lower", "left", "right", and the like are used merely to indicate relative positional relationships, and when the absolute position of the object being described is changed, the relative positional relationships may also be changed accordingly.

Examples

Referring to fig. 1 and fig. 2, the map lightweight compression transmission method in this embodiment includes a map compression step at the Linux server and a map decompression step at the Windows client, where the map compression step at the Linux server includes:

step S101, a 2D grid map established at a certain moment is obtained.

As shown in fig. 3, the 2D grid map established at a certain time refers to a map established at a certain time in the process of establishing a map, and it can be seen from fig. 3 that the 2D grid map mainly includes: white pixels-free traffic areas, black pixels-obstacle areas, gray pixels-unknown areas.

In step S102, an overall size S of the 2D grid map is obtained as [ W, H ], where W represents a width of the map and H represents a length of the map.

Step S103, acquiring an inner contour of the 2D grid map, which specifically comprises the following steps:

and S1031, carrying out binarization processing on the 2D grid map. This step extracts black pixels from the original map, facilitating the next operation.

S1032, isolated point filtering processing of the 2D grid map. In the step, isolated and outlier black pixel points are filtered from the binaryzation black pixel points. The filtering of the isolated black pixel points is to use a convolution operator (3 multiplied by 3 neighborhood) to traverse each pixel point, and then to delete the black pixel points meeting the unsatisfied requirements according to the area control threshold of the isolated points, thereby achieving the purpose of deleting the isolated black pixel points.

S1033: and (4) rectangularizing black pixel points of the 2D grid map. This step is to perform the rectangularization of the black pixel points in the map filtered in step S104, that is, to expand 0.5 pixels from top to bottom, left to right, and to form a rectangle of 1 × 1 pixels with the black pixel as the center.

S1034, performing Boolean union operation on all the rectangles to obtain a plurality of polygonal outlines. In this step, the black pixels that are rectangularly formed in step S105 are forward biased, that is, each rectangle is expanded outward by 1 pixel to form a 3 × 3 rectangle, and all the expanded and biased rectangles are subjected to boolean operation to finally form a plurality of polygonal outlines.

S1035: and traversing each polygon contour to obtain an inner contour. This step traverses the polygon outline generated in step S106, and if the vertices of one polygon outline all fall within another polygon, the polygon outline is an inner outline (no polygon outline nesting of more than two layers occurs in the 2D grid map), thereby obtaining the polygon inner outline, see fig. 7.

Step S104, acquiring black pixels of the 2D grid map, specifically comprising:

and S1041, selectively filtering isolated points of the 2D grid map. Referring to fig. 6, in this step, isolated and outlier black pixels are filtered out from the binarized black pixels. The filtering of the isolated black pixel points is to use a convolution operator (3 multiplied by 3 neighborhood) to traverse each pixel point, then to screen out the deleted black pixel list L which does not meet the requirement according to the area control threshold of the isolated point, to traverse the list L, if the number of the white pixel points in the 3 multiplied by 3 neighborhood is less than or equal to 1, the deleted black pixel list L is deleted, otherwise, the deletion is not deleted, the purpose of selectively deleting the isolated black pixel points is achieved, and the black pixel points at the boundary are prevented from being deleted by mistake.

S1042, black pixels of the 2D grid map are extracted. In this step, the black pixel points filtered in step 108 are obtained, so that the purpose of extracting the black pixels of the 2D grid map is achieved.

Step S105, obtaining the outer contour of the 2D grid map, which specifically comprises the following steps:

s1051, a binarization process is performed on the 2D grid map.

And S1052, tracing the boundary of the binary map.

And S1053, filtering and deleting isolated lines of the boundary. The boundary acquired in step S1052 has an isolated line, which means that the boundary line has a long and narrow line shape, and as shown in fig. 4, it needs to be removed. The contour exhibits overlapping portions at isolated lines, i.e., on isolated lines, contour points have repeating portions. According to the characteristic, the coordinate points of the boundary are traversed, all repeated coordinate points are deleted, however, the situation that an endpoint is not removed, namely a sharp point, occurs on one side of the boundary where the isolated line points to, an angle formed by three coordinates on the boundary including the sharp point is smaller than a certain threshold, and the threshold is 10 degrees, so that the effect of removing the sharp point is achieved. This completes the filtering deletion of the boundary line.

S1054, acquiring the outline of the 2D grid map, and the extracting and acquiring step S1053 filters out the boundary of the isolated line, see fig. 5.

S1055, the outer contour, inner contour, black pixels and overall size of the 2D grid map are output, see fig. 8. And at this moment, the map compression of the Linux server is finished.

The map decompression step of the Windows client comprises

Step S106, acquiring the outer contour, the inner contour, the black pixels and the overall size of the 2D grid map. This step receives the outer contour, inner contour, black pixels and overall size of the 2D grid map transmitted through the network.

In step S107, a blank gray map is extracted from the overall size of the 2D grid map. In the step, the overall size of the 2D grid map is obtained, and a blank gray map is generated as an initial map according to the overall size.

In step S108, white pixels are filled in the outer contour. In this step, white pixels are filled in the outer contour on the initial blank gray map generated in step S107 using the outer contour of the 2D grid map received in step S106.

Step S109 fills gray pixels inside the inner contour. This step performs white pixel filling inside the outer contour on the map obtained in step S108, using the inner contour of the 2D grid map received in step S201.

In step S110, black pixels are filled on the basis of step S109. This step fills black pixels on the map obtained in step S109 with the black pixels of the 2D grid map received in step S201.

Step S111, decompress the 2D grid map. The step is to complete the 2D raster map task after decompression and reconstruction and complete the map decompression of the Windows client side for the output of the step. The effect diagram is shown in fig. 9.

Typical embodiments of the present invention are as follows:

1. map compression and reconstruction testing.

Hardware platform and language

And (3) testing environment: intel (R) core (TM) i7-10700K CPU @3.80GHz and 32GB internal memory;

test language: python language, and Cython optimization is used.

The map test effect is as follows:

as shown in fig. 3 and 9, it can be seen that the original map and the reconstructed map are consistent on the whole, the visualization effect is not affected, and compared with the method of directly transmitting the original map, the method greatly reduces the network resources occupied by map transmission.

And (3) map testing time:

the test results are shown in the following table, and the test map 1 is fig. 3, and the reconstruction effect graph thereof is fig. 9; the test map 2 is fig. 10(a), and the reconstruction effect map thereof is fig. 10 (b); the test map 3 is fig. 11(a), and the reconstruction effect map thereof is fig. 11 (b). The running time of the visible algorithm is relatively fast, and the real-time requirement of million-scale map compression and decompression is basically met.

Algorithm running time statistical table

Note: the above runs were performed 10 times per map.

Claims

1. The map lightweight compression transmission method is characterized by comprising a map compression step of a Linux server side and a map decompression step of a Windows client side, wherein the map compression step of the Linux server side comprises the following steps:

the step of acquiring the outer contour of the 2D grid map includes:

2-1) carrying out binarization processing on the 2D grid map;

2-2) carrying out boundary tracking on the binarized 2D grid map;

2-3) deleting the isolated line after the boundary tracking to obtain an outer contour;

the step of acquiring the inner contour of the 2D grid map includes:

2-4) carrying out binarization processing on the 2D grid map;

2-5) carrying out filtering processing on isolated points of the 2D grid map;

2-6) rectangularizing black pixel points in the 2D grid map;

2-7) performing Boolean operation on the rectangular black pixel points to obtain a plurality of polygonal outlines;

2-8) traversing each polygon outline to obtain an inner outline;

the map decompression step of the Windows client comprises the following steps:

7) and obtaining the decompressed 2D grid map.

2. The map lightweight compression transmission method according to claim 1, wherein in step 2-2), the boundary obtained after the boundary tracking includes the following information:

(I) coordinate information of the corresponding vertex;

(II) the order of the positions of the corresponding vertices;

(III) the corresponding vertices are arranged clockwise.

3. The map lightweight compression transmission method according to claim 1, wherein in step 2-3), the boundary obtained after the isolated line is deleted includes the following information:

(I) coordinate information of the corresponding vertex;

(II) the order of the positions of the corresponding vertices;

(III) the corresponding vertices are arranged clockwise.

4. The map lightweight compression transmission method according to claim 1, wherein in step 2-6), after the black pixels in the 2D grid map are rectangularly formed, the remaining rectangles include the following information:

(I) vertex coordinates of the rectangle;

(II) storing the vertices of the rectangle clockwise.

5. The map lightweight compression transmission method according to claim 1, wherein the step 2) of extracting black pixels of the 2D grid map includes:

2-9) carrying out selective filtering processing on isolated points of the 2D grid map;

2-10) obtaining black pixels of the 2D grid map.

6. The map lightweight compression transmission method according to claim 5, wherein in step 2-9), the isolated points are one or more black pixels, the area of the isolated points is smaller than a threshold, and no black pixel of other black pixels exists within a certain threshold distance from the periphery of the isolated points, and the black pixels of non-composition boundaries are selectively deleted.