CN-121767759-B - Self-adaptive cleaning control method for cleaning robot based on image texture analysis

Abstract

The invention belongs to the technical field of image analysis, and particularly relates to a self-adaptive cleaning control method of a cleaning robot based on image texture analysis, which comprises the steps of performing top hat transformation on an original image by utilizing multi-angle linear structural elements to obtain an anisotropic residual texture map and calculating a gray scale run matrix; the method comprises the steps of constructing gray scale travel Cheng Nixiang coupling inertia based on a gray scale and travel negative correlation mapping relation, obtaining morphological erosion residual ratio representing stain stability through morphological erosion operation, further fusing to generate sediment consolidation modulus, and adjusting the command advancing speed and command shearing moment of the robot according to the obtained ratio. The invention solves the problems of grid line interference and reflection misjudgment through the deep coupling of visual characteristics and physical stability, and realizes the accurate matching of cleaning strength and stain state.

Inventors

• WEI YIMING
• LU CUNHAO

Assignees

• 星逻智能科技(苏州)有限公司

Dates

Publication Date

20260508

Application Date

20260302

Claims (6)

1. 1. The self-adaptive cleaning control method of the cleaning robot based on the image texture analysis is characterized by comprising the following steps: acquiring an original image of a photovoltaic panel, processing the original image by utilizing linear structural elements with a plurality of angles, generating an anisotropic residual texture map from which a directional background is stripped, and calculating a gray scale run matrix; based on the negative correlation mapping relation between the run length and the gray level in the gray level run matrix, calculating the coupling inertia of the gray level run Cheng Nixiang to extract the hardening stain characteristics corresponding to the dark color long communication region, wherein the coupling inertia of the gray level run Cheng Nixiang The method comprises the following steps: ; is the total number of gray levels; Is the maximum possible run length; the values of the elements in the gray scale run matrix; Is the quantized gray level; Is the run length; Is zero-proof constant; And Adjusting an index for the coupling; performing multiple morphological erosion operations on the anisotropic residual texture map by using the disc structural elements, and obtaining a morphological erosion residual ratio representing the stability of the physical structure of the stain according to the ratio of the gray scale of the eroded image after erosion to the gray scale of the anisotropic residual texture map before erosion; the morphology erosion residual ratio is amplified in a nonlinear way and is fused with the coupling inertia of the gray scale stream Cheng Nixiang, so that sediment consolidation modulus for quantifying the mechanical peel strength of stains is obtained, and the sediment consolidation modulus is as follows: ; represents the sediment consolidation modulus; Representing the morphological erosion residual ratio; Is a normalized coefficient; Is a residual sensitive factor; an exponential function that is based on a natural constant; Calculating an instruction advancing speed and an instruction shearing moment based on the sediment consolidation modulus, and realizing self-adaptive adjustment of cleaning strength; Command travel speed The method comprises the following steps: ; the reference speed of the robot; as a hyperbolic tangent function; Is a speed regulation coefficient; The commanded shear torque is: ; Is the command shear torque; The no-load moment of the motor is set; Is a moment gain coefficient; Is a natural logarithmic function; Is a moment sensitivity coefficient.
2. 2. The method for adaptively cleaning and controlling the cleaning robot based on the image texture analysis according to claim 1, wherein the process for obtaining the morphological erosion residual ratio comprises the steps of: counting the gray value sum of all pixel points in the erosion image to be used as a residual pixel sum; counting the gray value sum of all pixel points in the anisotropic residual texture map to be used as a reference pixel sum; and calculating the ratio of the residual pixel sum to the reference pixel sum to obtain the morphological erosion residual ratio.
3. 3. The method for adaptively cleaning and controlling a cleaning robot based on image texture analysis according to claim 1, wherein the generating an anisotropic residual texture map from which a directional background is peeled off comprises: And respectively executing morphological open operation on the original image by utilizing linear structural elements with a plurality of angles, calculating a top hat image, taking minimum values of a plurality of top hat images at each space coordinate pixel point, and generating an anisotropic residual texture image from which a directional background is stripped.
4. 4. The adaptive cleaning control method for a cleaning robot based on image texture analysis according to claim 3, wherein the top hat map obtaining method comprises the steps of: and subtracting the gray value of the corresponding coordinate pixel point in the open operation image under each angle from the gray value of each pixel point in the original image to obtain residual components of each pixel point, wherein the residual components of all the pixel points form a top hat graph under the corresponding angle.
5. 5. The method for adaptively cleaning and controlling a cleaning robot based on image texture analysis according to claim 1, wherein the calculating a gray scale run matrix comprises: carrying out gray level equidistant quantization on the anisotropic residual texture map to obtain a 16-level gray level map; and counting and accumulating the runs according to four preset directions to generate a gray scale run matrix.
6. 6. The method of claim 1, wherein the plurality of angles includes 0 degrees, 45 degrees, 90 degrees, 135 degrees.

Description

Self-adaptive cleaning control method for cleaning robot based on image texture analysis Technical Field The invention relates to the technical field of image analysis. More particularly, the present invention relates to a cleaning robot adaptive cleaning control method based on image texture analysis. Background Photovoltaic power generation is an important component of renewable energy sources, and the cleanliness of the surface of a battery plate directly influences the energy conversion efficiency of a photovoltaic module. In outdoor operating environments, the surface of photovoltaic panels is extremely prone to accumulate various deposited stains containing not only loose dust but also stubborn board structures such as bird droppings, mud spots, or stucco that dry out for a long period of time. In order to maintain the power generation efficiency, currently, a cleaning robot is generally adopted in the industry to replace manual work to perform automatic operation, and the core premise of realizing self-adaptive adjustment of cleaning strength is that the stain state of the surface of the panel can be accurately monitored through a visual sensor. In the prior art, the dirt degree is usually estimated by using an image processing means, and the dirt distribution is mainly characterized by counting indexes such as long run dominance factors or gray non-uniformity in a gray run matrix. However, such standard statistical indicators have significant physical limitations in complex photovoltaic scenarios. Firstly, the photovoltaic panel itself has dense metal grid line background and cell dark lines, which overlap with deposited stains on the gray distribution, resulting in standard texture indexes that are very prone to misjudging regular grid line shadows as connected hardening stains. Secondly, the existing visual indexes often take color information and geometric connectivity as independent variables, lack deep characterization of physical compactness of stains, and cannot identify impurities with similar visual characteristics and different stripping difficulties, so that the robot cannot pertinently adjust output parameters of a mechanical actuating mechanism. The core technical problem that produces is that current detection method can't be under complicated photovoltaic panel background interference, accurate discernment spot physical consolidation intensity and turn into effective mechanical control logic. The cleaning robot is caused to face the technical contradiction that the cleaning robot causes power consumption and battery piece damage to common floating dust excessive cleaning or causes hot spot effect to the panel due to insufficient cleaning force to stubborn plate matters in actual operation. Disclosure of Invention The invention provides a self-adaptive cleaning control method of a cleaning robot based on image texture analysis, which solves the technical problem that the prior art can not accurately identify the soil consolidation strength and realize accurate mechanical control under the background of a complex photovoltaic panel; the method comprises the steps of calculating gray scale run Cheng Nixiang coupling inertia based on a negative correlation mapping relation between run length and gray scale in a gray scale run matrix to extract hardening stain characteristics corresponding to a dark color long communication area, performing repeated morphological corrosion operation on the anisotropic residual texture map by using disc structural elements, obtaining morphological corrosion residual ratio representing stability of a stain physical structure according to a ratio of corroded corrosion image to gray scale of the anisotropic residual texture map before corrosion, carrying out nonlinear amplification on the morphological corrosion residual ratio, and carrying out coupling inertia fusion with the gray scale run Cheng Nixiang to obtain sediment consolidation modulus for quantifying mechanical peel strength of the stain, and calculating instruction advancing speed and instruction shearing moment based on the sediment consolidation modulus to realize self-adaptive adjustment of cleaning strength. The method successfully establishes the association of the color shade and the communication length through the gray level game Cheng Nixiang coupling inertia, utilizes the physical characteristic that heavy stains are necessary to present a low gray level value on the photovoltaic panel due to light blocking, solves the problem that the shadows of the long strip grid lines are easy to confuse with the solid stains in the prior art, remarkably improves the accuracy of target identification, simultaneously introduces a stain ablation process under the action of mechanical force in which morphological erosion residual ratio is preformed in an algorithm, enables a robot to accurately distinguish impurities with similar visual characteristics and different physical properties, fills in the mapp