CN-116681779-B - Deep loosening disturbance surface extraction device based on binocular vision

CN116681779BCN 116681779 BCN116681779 BCN 116681779BCN-116681779-B

Abstract

The invention relates to a deep loosening disturbance surface extraction device based on binocular vision. The device acquires camera internal parameters by adopting Zhang Zhengyou calibration method, finishes camera calibration by shooting a plurality of plane target images under different angles, calculates depth information of deep loosening disturbance surface images by utilizing SGBM algorithm, generates an initial parallax map by selecting parallaxes of corresponding pixel points of left and right eye images, constructs a global energy function, solves optimal parallaxes by minimizing the energy function, and realizes accurate matching of binocular images. In the implementation process, the disturbance soil after subsoiling is removed by adopting a manual soil digging mode, so that the subsoiling ditch bottom line is exposed, the acquisition device is moved to the position above the ditch bottom line, the rack is kept horizontal by means of level adjustment, and the depth camera D435i is controlled by a motor to slowly and stably move along the track, so that the image acquisition of the whole unidirectional stroke is completed. The invention can efficiently and accurately extract the information of the subsoiler disturbance surface, and is suitable for the detection and evaluation of the quality of agricultural subsoiler operation.

Inventors

LI XIA
You Birong
JIANG ZHANGJUN

Assignees

天津理工大学

Dates

Publication Date: 20260512
Application Date: 20230607

Claims (9)

1. The utility model provides a dark pine disturbance face extraction element based on binocular vision, includes spirit level (1), camera platform (2), guide rail (3), slider (4), binocular vision camera (5) and frame (6), and its process of extracting dark pine disturbance face is as follows: s1, obtaining camera internal parameters by using Zhang Zhengyou calibration method, shooting images of a plurality of plane targets by using cameras under different angles, and performing camera calibration by performing calculation and analysis on corner points of a checkerboard S2, calculating depth of the deep scarification disturbance surface image by using an SGBM algorithm, selecting parallax of each pixel point at corresponding positions of left and right eye images, generating an initial parallax image, setting a global energy function, minimizing the energy function, and solving out optimal parallax as a binocular image matching representation S3, throwing out the soil disturbed after subsoiling by adopting a manual soil digging method to expose the bottom line of the subsoiling ditch S4, moving the device to the position above the ditch bottom line from which soil is thrown, and keeping the machine frame horizontal according to the position of the level meter And S5, slowly and stably advancing the depth camera D435i along the track by controlling a motor REALSENSE until the whole one-way stroke is finished.
2. The deep loosening disturbance surface extraction device based on binocular vision according to claim 1, wherein the binocular vision camera is connected with a camera platform, the camera is guaranteed to be parallel to the platform, and the shooting direction of the camera is perpendicular to a horizontal plane.
3. The deep scarification disturbance surface extraction device based on binocular vision according to claim 1, wherein the calibration of the binocular vision camera in step S1 comprises the steps of: step S11 obtains a mapping matrix H for each image Step S12, solving the internal reference matrix A linearly by using constraint conditions Step S13 of maximum likelihood estimation The device is calibrated by adopting Zhang Zhengyou calibration method, and tangential distortion is not considered.
4. The deep scarification disturbance surface extraction device based on binocular vision according to claim 1, wherein the step S2 of calculating the image depth by using the SGBM algorithm comprises the steps of: step S21 is based on depth calculation of binocular images, and the basic principle is that three-dimensional scene coordinate information is calculated by combining parameters obtained by calibrating binocular cameras through pixel level differences of corresponding point positions of the same point in a scene in left and right images Step S22, calculating the parallax of each pixel point at the corresponding position of the left eye image and the right eye image by using an energy function, generating an initial parallax map, setting a global energy function, minimizing the energy function, and solving the optimal parallax as a binocular image matching representation.
5. The deep loosening disturbance surface extraction device based on binocular vision according to claim 1, wherein the disturbed soil is thrown out manually in the step S3, and no force is generated on the undisturbed soil in the process.
6. The deep scarification disturbance surface extraction device based on binocular vision according to claim 1, wherein the deep scarification disturbance surface extraction device is characterized in that in the step S4, when the deep scarification condition is unchanged by default, the ditch bottom line shape is uniform.
7. The deep loosening disturbance surface extraction device based on binocular vision according to claim 1, wherein in the step S4, the position of the whole structure is ensured by a level meter so as to avoid measurement errors caused by poor flatness of farmland soil.
8. The deep loosening disturbance surface extraction device based on binocular vision according to claim 1, wherein in the step S5, the slider is advanced to complete a unidirectional stroke to extract the shape of the whole ditch bottom line, and the slider does not need to return after reaching one end.
9. The binocular vision-based deep scarification surface extraction apparatus of claim 1, wherein the slider is slowly advanced at a steady speed during the advance in step S5, so as to avoid missing data points due to too high speed.

Description

Deep loosening disturbance surface extraction device based on binocular vision Technical Field The invention relates to the technical field of soil deep scarification, in particular to a deep scarification disturbance surface extraction device based on binocular vision. Background Subsoiling is used as one of important ways of soil protective cultivation, a cultivation layer can be deepened, a plough bottom layer is broken, the shape of a disturbance surface is an important factor affecting crop growth and comprehensive subsoiling effects in the process of subsoiling among crop cultivation rows and comprehensive subsoiling of the soil, and the current method for determining the area of the subsoiling disturbance surface mainly comprises the steps of manually selecting a plurality of cross section points on a subsoiling ditch bottom line for grooving after subsoiling operation, and then measuring the shape of the subsoiling ditch bottom line by using a plugboard method or a laser method. However, the plugboard method has lower efficiency and poorer precision, and particularly wastes a great deal of time and manpower and material resources when a great deal of data needs to be acquired, while the laser method has greatly influenced the measurement precision when the illumination intensity is higher and has higher manufacturing cost. Binocular stereo vision is a method for acquiring three-dimensional geometric information of an object to be measured from different positions by using imaging equipment based on parallax principle, and REALSENSE depth cameras can rapidly extract and output distances between the cameras and a certain point to obtain a series of coordinate values of the shot points, and the method is low in price and less affected by environment. In view of this, it is necessary to design a method for applying binocular stereoscopic vision technology to deep scarification of surface ditch bottom line extraction. Disclosure of Invention Aiming at the problems, the method designs a device for rapidly extracting the deep scarification disturbance surface by using a binocular vision camera. The invention aims at realizing the following technical scheme: the deep loosening disturbance surface extraction device based on binocular vision mainly comprises a rack, a sliding rail, REALSENSE depth cameras D534i, a camera platform and a level meter. The sliding rail is connected with the binocular vision camera through the camera platform, so that the camera can stably move in the linear direction, the influence of frame dropping of the binocular vision camera on a data result is avoided, and the accuracy of the obtained three-dimensional coordinates is improved. The binocular vision camera is installed below the platform, is parallel to the ground when working, and the camera shoots soil right below. The level gauge is arranged above the frame so as to adjust the position of the whole machine in the working process, so that the level gauge is parallel to the ground, and measurement errors caused by poor soil flatness are avoided. In the above technical solution, the disturbance surface extraction method includes the following steps: Step S1, obtaining camera internal parameters by using Zhang Zhengyou calibration method, shooting images of a plurality of plane targets such as checkerboard images by using the camera under different angles, and then performing camera calibration (namely solving internal and external parameters of the camera) by performing calculation and analysis on corner points of the checkerboard. And S2, calculating depth of the deep scarification disturbance surface image by using an SGBM algorithm, selecting parallax of each pixel point at corresponding positions of the left eye image and the right eye image, generating an initial parallax map, setting a global energy function, minimizing the energy function, and solving the optimal parallax as a binocular image matching representation. And S3, manually throwing out soil above the trench bottom line. And S4, moving the machine to the position above the ditch bottom line from which soil is thrown, and keeping the machine frame horizontal according to the position adjustment of the level meter. And S5, slowly and stably advancing the depth camera D534i along the track by controlling the motor REALSENSE until the whole one-way stroke is finished. Specifically, in step S1, the calibration of the binocular vision camera includes the steps of: step S11 obtains a mapping matrix (homography matrix) H for each image. One two-dimensional point is represented by m= (u, v) T, one three-dimensional point can be represented by m= (X, Y, Z) T, and the augmentation matrix isAndThe relationship between a three-dimensional point and its projected image point is: In the formula, s is any standard vector, A is matrix internal reference, R (rotation matrix) and t (translation vector) are external reference. Internal parameters: Wherein (u 0,v0) is the princip