CN-121994716-A - Liquid grease optical characteristic on-line measuring system based on computer vision

CN121994716ACN 121994716 ACN121994716 ACN 121994716ACN-121994716-A

Abstract

The invention relates to the technical field of grease quality detection, and discloses a liquid grease optical characteristic online detection system based on computer vision. The system acquires polarized image data through multi-angle polarization imaging, calculates Stokes parameters to generate a polarized characteristic image, applies an anisotropic Gaussian filter bank to combine consistency judgment to detect fibrous foreign matters and generate a spatial distribution mask, applies shearing disturbance pulses to trigger grease structure damage, acquires time sequence flow velocity field data in a recovery process through structured light phase demodulation to generate a spatial resolution viscosity recovery time chart sequence, divides regions based on the fibrous spatial distribution mask and extracts recovery parameters, finally generates a quantitative evaluation result of fiber-rheological coupling, and realizes quantitative association of the spatial distribution of the fibrous foreign matters and local thixotropic recovery.

Inventors

CUI RUIFU
WANG YIFAN
XIN WEI
FAN LINLIN
WEN LEI
WANG LEI
CAO HONGYING
Qie Wenbin
CUI LIANYING
SU JUAN
LIU YANWEI
CHEN XINGHAI

Assignees

瑞福油脂股份有限公司

Dates

Publication Date: 20260508
Application Date: 20260407

Claims (10)

1. The liquid grease optical characteristic on-line detection method based on computer vision is characterized by comprising the following steps of: Sequentially irradiating the flowing grease detection area under a plurality of polarization angles, synchronously collecting a plurality of corresponding polarized images, and obtaining multi-angle polarized image data; calculating Stokes parameters of each pixel position based on the multi-angle polarized image data to generate a polarized image and a polarized image; Extracting linear response intensity in each direction by applying an anisotropic Gaussian filter bank containing a plurality of elongated kernel functions in different directions to the polarization degree image, determining a fibrous foreign matter candidate region through consistency judgment by combining polarization angle values of corresponding positions in the polarization angle image, and performing skeletonizing treatment on the fibrous foreign matter candidate region to extract a central line to generate fiber space distribution mask and fiber geometric characteristic data; Applying a shear disturbance pulse upstream of a flowing grease detection area to trigger a damage process of a grease structure; Projecting a structured light pattern to a flowing grease detection area from the finishing moment of shearing disturbance, continuously acquiring a deformation image sequence at fixed time intervals, and acquiring time sequence flow velocity field data of a recovery process through phase demodulation; applying standard intensity detection pulse disturbance to each time point in the time sequence flow velocity field data of the recovery process, measuring flow velocity response characteristics of each spatial position, calculating an instantaneous viscosity value, and generating a viscosity recovery time chart sequence; Dividing a viscosity recovery time chart sequence into a fiber adjacent area and a fiber-free area based on a fiber space distribution mask, respectively applying exponential recovery model fitting to viscosity recovery curves of the fiber adjacent area and the fiber-free area, and extracting recovery time constants and recovery degree parameters of the fiber adjacent area and the fiber-free area; And calculating the ratio of the recovery time constant of the fiber adjacent area to the recovery time constant of the fiber-free area and the spatial distribution difference statistic to generate a spatial coupling quantitative evaluation result of the influence of the fiber on the thixotropic recovery.
2. The method for detecting the optical characteristics of the liquid grease on line based on computer vision according to claim 1 is characterized in that the polarization angles are four polarization angles of 0 °, 45 °,90 ° and 135 °, the stokes parameters comprise a total light intensity parameter, a horizontal and vertical polarization component difference parameter and a 45 ° and 135 ° polarization component difference parameter, the polarization degree is calculated according to the ratio of the square root of the sum of the square of the horizontal and vertical polarization component difference parameter and the square of the 45 ° and 135 ° polarization component difference parameter to the total light intensity parameter, and the polarization angle is calculated according to one half of the arctangent value of the ratio of the 45 ° and 135 ° polarization component difference parameter and the horizontal and vertical polarization component difference parameter.
3. The method for detecting the optical characteristics of the liquid grease on line based on the computer vision according to claim 1 is characterized in that the anisotropic Gaussian filter bank comprises M elongated kernel functions in different directions, the k-th elongated kernel function is defined as a two-dimensional Gaussian function extending along a direction angle, the direction angles are uniformly distributed according to the interval of 180 degrees divided by M, the elongated kernel function has a standard deviation along a long axis direction and a standard deviation along a short axis direction in a local coordinate system, the standard deviation along the long axis direction is larger than the standard deviation along the short axis direction, and convolution operation is carried out on the elongated kernel functions in all M directions on each pixel position in the polarization degree image to obtain filter response intensity values in the M directions.
4. The method for detecting the optical characteristics of the liquid grease on line based on the computer vision according to claim 3 is characterized in that the consistency judgment is carried out by determining a direction angle generating the maximum filter response for each pixel position in the polarization degree image, calculating an angle difference value between the direction angle and a polarization angle value at a corresponding position in the polarization angle image, and judging the pixel position as a fibrous foreign matter candidate region when the angle difference value is smaller than a consistency threshold value and the maximum filter response intensity is larger than a response intensity threshold value, wherein the response intensity threshold value is determined according to the average value of the response intensity of a non-fibrous background region in the polarization degree image plus three times of standard deviation.
5. The method for detecting the optical characteristics of the liquid grease based on the computer vision according to claim 1 is characterized in that the skeletonizing treatment adopts a morphological refinement algorithm, non-central line pixels of the fibrous foreign matter candidate region are removed through iterative erosion, and the connectivity of fibers is maintained, wherein the fiber geometric characteristic data comprises a central line coordinate sequence, a length, an average width and an orientation angle of each fiber segment, the fiber length is the cumulative arc length of the central line coordinate sequence, the average width is the average width of the fibrous foreign matter candidate region in the direction perpendicular to the central line, and the orientation angle is the weighted average value of tangential direction angles of the central line at each position.
6. The method for on-line detection of optical characteristics of liquid grease based on computer vision according to claim 1 is characterized in that the shear disturbance pulse is generated by a shear actuator located at the upstream of the flowing grease detection area, the shear actuator applies a shear action to the flowing grease at a preset shear rate within the pulse duration, the phase demodulation adopts a fourier transform method to perform two-dimensional fourier transform on the deformed image, extracts a frequency spectrum component corresponding to the fundamental frequency of the structured light in the frequency domain, filters and inverse fourier transforms the frequency spectrum component to obtain a complex form demodulation signal, and obtains a phase value by calculating the argument of the demodulation signal.
7. The method for on-line detection of optical characteristics of liquid grease based on computer vision according to claim 6 is characterized in that the phase change amount between adjacent time points is converted into a displacement amount, the displacement amount is equal to the phase change amount divided by 2π and multiplied by the space period of the structured light pattern, the flow velocity value of each space position is equal to the displacement amount divided by the acquisition time interval, and when the phase change amount exceeds the interval from minus pi to plus pi, the phase unwrapping algorithm is adopted for processing to eliminate phase ambiguity.
8. The method for detecting the optical characteristics of the liquid grease on line based on the computer vision according to claim 1 is characterized in that the shearing stress of the standard intensity detection pulse disturbance is 5-10% of the shearing stress of the shearing disturbance pulse, and the calculation process of the instantaneous viscosity value is that the flow velocity value before and after the standard intensity detection pulse disturbance is applied is recorded, the flow velocity response amplitude is calculated, the flow velocity response amplitude is divided by the characteristic size of a flow channel to obtain the shearing rate, and the shearing stress corresponding to the standard intensity detection pulse disturbance is divided by the shearing rate to obtain the instantaneous viscosity value.
9. The method for detecting the optical characteristics of the liquid grease based on the computer vision according to claim 1 is characterized in that the fiber adjacent area is defined as an area in which the fiber center line in the fiber space distribution mask expands outwards within a preset distance which is 2 times to 10 times of the average width of the fiber, the index restoration model represents an index function that the viscosity value is equal to the equilibrium viscosity value minus the difference between the equilibrium viscosity value and the initial viscosity value multiplied by the ratio of negative time to restoration time constant, the restoration degree parameter is defined as the difference between the equilibrium viscosity value and the initial viscosity value divided by the difference between the reference viscosity value and the initial viscosity value, and the space distribution difference statistic comprises the space variation coefficient ratio of the restoration time constant between the fiber adjacent area and the fiber-free area, the space average difference of the restoration degree parameter and the correlation coefficient of the fiber orientation angle and the local restoration time constant distribution.
10. A computer vision-based liquid grease optical characteristic online detection system for executing the computer vision-based liquid grease optical characteristic online detection method according to any one of claims 1 to 9, characterized by comprising: the multi-angle polarized image acquisition module is used for sequentially irradiating the flowing grease detection area under a plurality of polarized angles, synchronously acquiring a plurality of corresponding polarized images and acquiring multi-angle polarized image data; The polarization characteristic image generation module is used for calculating Stokes parameters of each pixel position based on the multi-angle polarization image data to generate a polarization degree image and a polarization angle image; The fibrous foreign matter detection module is used for extracting linear response intensity in all directions by applying an anisotropic Gaussian filter bank to the polarization degree image, determining fibrous foreign matter candidate areas through consistency judgment by combining the polarization angle image, and performing skeletonizing treatment to generate fiber space distribution masks and fiber geometric characteristic data; The shear disturbance applying module is used for applying a shear disturbance pulse at the upstream of the flowing grease detection area to trigger the damage process of the grease structure; The time sequence flow velocity field acquisition module is used for projecting a structured light pattern to the flowing grease detection area, continuously acquiring a deformed image sequence and acquiring time sequence flow velocity field data in a recovery process through phase demodulation; the viscosity recovery time chart generation module is used for applying standard intensity detection pulse disturbance to each time point in the time sequence flow velocity field data of the recovery process, measuring flow velocity response characteristics, calculating an instantaneous viscosity value and generating a viscosity recovery time chart sequence; The recovery parameter extraction module is used for dividing the viscosity recovery time chart sequence into a fiber adjacent area and a fiber-free area based on a fiber space distribution mask, and fitting and extracting recovery time constants and recovery degree parameters by using an index recovery model; and the coupling evaluation result generation module is used for calculating the ratio of the fiber adjacent area to the recovery time constant of the fiber-free area and the spatial distribution difference statistic to generate a spatial coupling quantitative evaluation result of the influence of the fiber on the thixotropic recovery.

Description

Liquid grease optical characteristic on-line measuring system based on computer vision Technical Field The invention relates to the technical field of grease quality detection, in particular to a liquid grease optical characteristic on-line detection system based on computer vision. Background Sesame oil can be combined with other ingredients (e.g., nanocellulose, sterols, etc.) to form an oleogel. For example, studies have shown that walnut oil can be converted to solid "margarine" by the addition of edible nanocellulose, a similar approach applies to sesame oil, forming a stable network structure by encapsulating the oil droplets. In the production process of the structured grease containing fibrous foreign matters, the fibrous foreign matters comprise filter material shedding fiber yarns, packaging material scraps and the like, and the structured grease comprises crystal network-containing shortening and colloid-containing blend oil thixotropic grease. The fiber is used as heterogeneous core to be embedded into the crystal or colloid network of grease, the long axis direction of the fiber tends to be consistent with the flowing direction of the grease due to the slender form of the high length-width ratio, the fiber presents obvious directional distribution characteristics, and the depolarization effect of the fiber presents anisotropy along the long axis and the short axis direction of the fiber. When the grease needs to recover its thixotropic structure after undergoing shear failure, the presence of fibers may accelerate or hinder the process of structural reconstruction in localized areas. In the prior art, the quality of the grease is evaluated by adopting a separate fiber detection method and a separate thixotropic measurement method. The individual fiber detection method typically employs a conventional blob detection algorithm to detect foreign matter in the grease. The thixotropic property of the grease was evaluated by measuring the global viscosity recovery parameter after the application of a shear disturbance by a separate thixotropic measurement method. The prior art has the following defects that an independent fiber detection method can only position fiber foreign matters and cannot evaluate the influence of the fiber foreign matters on the rheological property of grease, a traditional blob detection algorithm assumes isotropic round spots, has low detection sensitivity on slender fibers and cannot extract direction and length information, an independent thixotropic measurement method outputs global recovery parameters and cannot identify abnormal recovery behaviors of areas nearby the fibers, whether the grease is in a low-viscosity state after structural failure or a high-viscosity state after recovery cannot be distinguished, and quantitative association relation between fiber space distribution and local thixotropic recovery cannot be established when the two technologies are respectively executed, so that the actual influence of the fiber foreign matters on the rheological property of the thixotropic grease cannot be accurately evaluated. Disclosure of Invention The invention provides a liquid grease optical characteristic online detection system based on computer vision, which solves the technical problems that fibrous foreign matters in grease cannot be accurately identified and the influence of the fibrous foreign matters on a thixotropic recovery process cannot be quantitatively evaluated in the related technology. The invention discloses a liquid grease optical characteristic online detection method based on computer vision, which comprises the following steps of sequentially irradiating a flowing grease detection area under a plurality of polarization angles, synchronously collecting a plurality of corresponding polarized images to obtain multi-angle polarized image data, calculating Stokes parameters of each pixel position based on the multi-angle polarized image data to generate a polarization degree image and a polarization angle image, applying an anisotropic Gaussian filter bank containing a plurality of elongated kernel functions in different directions to the polarization degree image to extract linear response intensity in each direction, combining polarization angle values of corresponding positions in the polarization angle image to determine a fibrous foreign matter candidate area through consistency judgment, skeletonizing the fibrous foreign matter candidate area to extract a central line to generate a fiber space distribution mask and fiber geometric characteristic data, applying a shearing disturbance pulse on the upstream of the flowing grease detection area to trigger a damage process of a grease structure, starting from the shearing disturbance end moment, continuously collecting a deformation image sequence at a fixed time interval, obtaining a recovery process time sequence flow velocity field data through phase demodulation, applying standard intensity dete