Plastic mold injection monitoring system and method

Abstract

The present invention provides a plastic mold injection monitoring system and method thereof. The system includes an injection monitoring center, multiple machine vision camera modules, a first temperature recognition module, a second temperature recognition module, and an injection temperature monitoring module. The machine vision camera module collects a surface visible light image and an initial infrared thermal image of a target plastic mold. The first temperature recognition module performs feature analysis on the surface visible light image to obtain surface features of the plastic mold, and inputs the surface features of the plastic mold into a first temperature recognition model to obtain a first temperature result. The second temperature recognition module fuses the initial infrared thermal image to obtain a fused infrared thermal image, and inputs the fused infrared thermal image into a second temperature recognition model to obtain a second temperature result. The injection temperature monitoring module fuses the first and second temperature results to obtain final temperature information during the injection molding process. The present invention improves the accuracy of temperature monitoring of plastic molds during the injection molding process.

Description

Technical Field

[0001] The present invention relates to the technical field of computers, and in particular to a plastic mold injection monitoring system and method thereof. Background Art

[0002] During the plastic mold injection process, temperature monitoring is a key step in ensuring product quality. Currently, commonly used plastic mold injection temperature monitoring methods mainly include direct thermocouple measurement and infrared temperature measurement.

[0003] The direct thermocouple measurement method inserts a thermocouple probe directly into the mold or near the injection molding area, and measures changes in the electrical signal to reflect temperature changes. However, the installation location of the thermocouple is limited, making it impossible to accurately measure critical temperature change areas. In addition, the thermocouple probe is easily impacted and abraded by the material during the injection molding process, shortening its service life and thus reducing the temperature measurement accuracy, resulting in reduced temperature monitoring accuracy. The infrared temperature measurement method determines the temperature through non-contact measurement of infrared radiation on the mold surface. However, it is easily interfered with by external environmental factors such as ambient temperature, humidity, and light, which can affect the accuracy of temperature measurement and reduce the accuracy of temperature monitoring. Summary of the Invention

[0004] The present invention provides a plastic mold injection monitoring system and method thereof, aiming to improve temperature measurement accuracy and improve the accuracy of temperature monitoring of the plastic mold during the injection molding process.

[0005] In a first aspect, the present invention provides a plastic mold injection monitoring system, comprising an injection monitoring center, multiple machine vision camera modules, a first temperature recognition module, a second temperature recognition module, and an injection temperature monitoring module; the multiple machine vision camera modules are installed above and on the side of the injection area of ​​the plastic mold, and the multiple machine vision camera modules include a visible light camera module and an infrared camera module; the injection monitoring center is respectively connected to the multiple machine vision camera modules, the first temperature recognition module, the second temperature recognition module, and the injection temperature monitoring module to manage each module;

[0006] A plurality of the machine vision camera modules are used to capture a surface visible light image and a plurality of initial infrared thermal images of a target plastic mold;

[0007] The first temperature recognition module is configured to perform feature analysis on the surface visible light image to obtain surface features of the target plastic mold, and input the surface features into a first temperature recognition model to obtain a first temperature result output by the first temperature recognition model;

[0008] The second temperature recognition module is configured to fuse the multiple initial infrared thermal images to obtain a fused infrared thermal image, and input the fused infrared thermal image into a second temperature recognition model to obtain a second temperature result output by the second temperature recognition model;

[0009] The injection molding temperature monitoring module is configured to obtain final temperature information of the target plastic mold during the injection molding process by fusing the first temperature result and the second temperature result;

[0010] The first temperature recognition model is trained based on the surface features of the sample and its corresponding first temperature result label; the second temperature recognition model is trained based on the infrared thermal image of the sample and its corresponding second temperature result label.

[0011] In a second aspect, the present invention further provides a plastic mold injection monitoring method, which is implemented based on the plastic mold injection monitoring system described in the first aspect. The plastic mold injection monitoring method includes:

[0012] A plurality of machine vision cameras installed above and on the side of the injection area of ​​the plastic mold are used to collect a surface visible light image and a plurality of initial infrared thermal images of the target plastic mold;

[0013] performing feature analysis on the surface visible light image to obtain surface features of the target plastic mold, and inputting the surface features into a first temperature recognition model to obtain a first temperature result output by the first temperature recognition model; the first temperature recognition model is trained based on the sample surface features and their corresponding first temperature result labels;

[0014] fusing the plurality of initial infrared thermal images to obtain a fused infrared thermal image, and inputting the fused infrared thermal image into a second temperature recognition model to obtain a second temperature result output by the second temperature recognition model; the second temperature recognition model is trained based on sample infrared thermal images and their corresponding second temperature result labels;

[0015] Based on the fusion of the first temperature result and the second temperature result, final temperature information of the target plastic mold during the injection molding process is obtained.

[0016] In a third aspect, the present invention further provides an electronic device comprising: a memory for storing a computer software program; and a processor for reading and executing the computer software program, thereby implementing any of the above-described methods for monitoring plastic mold injection.

[0017] In a fourth aspect, the present invention further provides a non-transitory computer-readable storage medium, wherein the storage medium stores a computer software program, and when the computer software program is executed by a processor, it implements any of the above-mentioned plastic mold injection monitoring methods.

[0018] In a fifth aspect, the present invention further provides a computer program product, comprising a computer program, which, when executed by a processor, implements any of the above-described methods for monitoring plastic mold injection.

[0019] The plastic mold injection monitoring system provided by the embodiment of the present invention can fully cover the injection mold from different angles and positions by setting multiple machine vision-based visible light cameras and infrared cameras above and on the sides of the injection area, and can obtain a large amount of mold surface information, thereby analyzing a more comprehensive and detailed temperature distribution, and can capture local subtle temperature differences. On the other hand, the temperature information of the surface visible light image and the infrared thermal image are predicted separately through the temperature recognition model, and then the temperature information of the surface visible light image and the infrared thermal image are fused to obtain the final temperature information of the plastic mold during the injection molding process, thereby reducing temperature errors. Therefore, the embodiment of the present invention improves the accuracy and stability of temperature measurement, achieves improved temperature measurement accuracy, and improves the accuracy of temperature monitoring of plastic molds during the injection molding process. BRIEF DESCRIPTION OF THE DRAWINGS

[0020] Figure 1 It is a structural diagram of the plastic mold injection monitoring system provided by the present invention;

[0021] Figure 2 1 is a flow chart of the plastic mold injection monitoring method provided by the present invention;

[0022] Figure 3 An embodiment diagram of an electronic device provided by an embodiment of the present invention;

[0023] Figure 4 An embodiment diagram of a computer-readable storage medium provided for an embodiment of the present invention. DETAILED DESCRIPTION

[0024] The following will clearly and completely describe the technical solutions in the embodiments of the present invention in conjunction with the accompanying drawings. Obviously, the described embodiments are only part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without making any creative efforts shall fall within the scope of protection of the present invention.

[0025] In the description of the present invention, the terms "first" and "second" are used for descriptive purposes only and should not be understood to indicate or imply relative importance or implicitly specify the number of the technical features indicated. Therefore, a feature specified as "first" or "second" may explicitly or implicitly include one or more of the specified features. In the description of the present invention, "plurality" means two or more, unless otherwise specifically defined.

[0026] In the description of the present invention, the term "for example" is used to mean "used as an example, illustration or illustration". Any embodiment of the present invention described as "for example" is not necessarily to be construed as being more preferred or advantageous than other embodiments. The following description is given to enable any person skilled in the art to implement and use the present invention. In the following description, details are listed for the purpose of explanation. It should be understood that a person of ordinary skill in the art can recognize that the present invention can be implemented without using these specific details. In other examples, well-known structures and processes are not elaborated in detail to avoid obscuring the description of the present invention with unnecessary details. Therefore, the present invention is not intended to be limited to the embodiments shown, but is consistent with the widest scope consistent with the principles and features disclosed herein.

[0027] Optional, see Figure 1 As shown, Figure 1 It is a structural diagram of the plastic mold injection monitoring system provided by the present invention. The plastic mold injection monitoring system includes an injection monitoring center, multiple machine vision camera modules, a first temperature recognition module, a second temperature recognition module and an injection temperature monitoring module. Among them, the machine vision camera module in the embodiment of the present invention can be understood as a machine vision camera.

[0028] Optionally, multiple machine vision camera modules in an embodiment of the present invention are installed above and on the sides of the injection area of ​​the plastic mold. The multiple machine vision camera modules include a visible light camera module and multiple infrared camera modules. The injection molding monitoring center is respectively connected to the multiple machine vision camera modules, the first temperature recognition module, the second temperature recognition module and the injection molding temperature monitoring module, so that each module can be managed.

[0029] Optionally, the visible light camera module among the multiple machine vision camera modules captures a visible light image of the surface of the target plastic mold, and the multiple infrared camera modules capture multiple initial infrared thermal images of the target plastic mold.

[0030] Optionally, the first temperature recognition module performs feature analysis on the surface visible light image to obtain the plastic mold surface features of the target plastic mold.

[0031] Furthermore, the first temperature recognition module inputs the surface features of the plastic mold into the first temperature recognition model to obtain a first temperature result output by the first temperature recognition model, wherein the first temperature recognition model in the embodiment of the present invention is trained based on the sample surface features and their corresponding first temperature result labels.

[0032] Optionally, the second temperature recognition module fuses multiple initial infrared thermal images to obtain a fused infrared thermal image.

[0033] Furthermore, the second temperature recognition module inputs the fused infrared thermal image into the second temperature recognition model to obtain a second temperature result output by the second temperature recognition model, wherein the second temperature recognition model in the embodiment of the present invention is trained based on the sample infrared thermal image and its corresponding second temperature result label.

[0034] Optionally, the injection temperature monitoring module fuses the first temperature result and the second temperature result to obtain final temperature information of the target plastic mold during the injection molding process.

[0035] By setting up multiple machine vision-based visible light cameras and infrared cameras above and on the sides of the injection molding area, the embodiment of the present invention can fully cover the injection mold from different angles and positions, obtain a large amount of mold surface information, and then analyze a more comprehensive and detailed temperature distribution, and capture local subtle temperature differences. On the other hand, the temperature information of the surface visible light image and infrared thermal image is predicted separately through the temperature recognition model, and then the temperature information of the surface visible light image and infrared thermal image are fused to obtain the final temperature information of the plastic mold during the injection molding process, thereby reducing temperature errors. Therefore, the embodiment of the present invention improves the accuracy and stability of temperature measurement, achieves improved temperature measurement precision, and improves the accuracy of temperature monitoring of the plastic mold during the injection molding process.

[0036] Optional, see Figure 2 , Figure 2 : is a flow chart of the plastic mold injection monitoring method provided by the present invention. In the embodiment of the present invention, the execution subject of the plastic mold injection monitoring method is the injection monitoring system. Therefore, the plastic mold injection monitoring method includes:

[0037] In step 10 , a plurality of machine vision cameras installed above and on the side of the injection area of ​​the plastic mold are used to collect a surface visible light image and a plurality of initial infrared thermal images of the target plastic mold.

[0038] Optionally, in an embodiment of the present invention, multiple machine vision cameras are installed above and on the side of the injection molding area of ​​the plastic mold. The multiple machine vision cameras include a visible light camera and multiple infrared cameras. Therefore, the injection molding monitoring system can call the visible light camera to collect the surface visible light image of the target plastic mold, and call the multiple infrared cameras to collect multiple initial infrared thermal images of the target plastic mold. In one embodiment, the response function of the camera is R(x, y, λ, t), where (x, y) represents the coordinates on the image plane, λ represents the wavelength of light, and t represents time. For visible light images, the wavelength range is λ vis For infrared thermal images, the wavelength range λ in Inside, the surface visible light image I is collected vis and multiple initial infrared thermal images I ir .

[0039] In step 20 , feature analysis is performed on the surface visible light image to obtain surface features of the target plastic mold, and the surface features of the target plastic mold are input into a first temperature recognition model to obtain a first temperature result output by the first temperature recognition model.

[0040] Optionally, the injection molding monitoring system performs feature analysis on the surface visible light image to obtain the plastic mold surface features of the target plastic mold, wherein the plastic mold surface features include texture, color, and glossiness. The specific analysis process is as described in steps 201 to 204.

[0041] Optionally, the injection molding monitoring system of an embodiment of the present invention is embedded with a first temperature recognition model, wherein the first temperature recognition model is trained based on the surface features of the sample and its corresponding first temperature result label, and texture-temperature, color-temperature, and glossiness-temperature mapping relationships are established in the first temperature recognition model.

[0042] Therefore, the injection molding monitoring system inputs the surface features of the plastic mold into the first temperature recognition model. The first temperature recognition model maps the surface features of the plastic mold according to the texture-temperature, color-temperature, and glossiness-temperature mapping relationships, and outputs the first temperature results corresponding to the surface features of the plastic mold, as specifically described in steps 205 to 208.

[0043] In step 30 , the multiple initial infrared thermal images are fused to obtain a fused infrared thermal image, and the fused infrared thermal image is input into a second temperature recognition model to obtain a second temperature result output by the second temperature recognition model.

[0044] Optionally, since different infrared cameras capture images at different positions and angles, the injection molding monitoring system needs to fuse multiple initial infrared thermal images to obtain a fused infrared thermal image, as specifically described in steps 301 to 303 .

[0045] Optionally, the injection molding monitoring system of this embodiment of the present invention includes a built-in second temperature recognition model, which is trained based on sample infrared thermal images and their corresponding second temperature result labels. Therefore, the injection molding monitoring system inputs the fused infrared thermal image into the second temperature recognition model, which then outputs the second temperature result.

[0046] Step 40 : A final temperature information of the target plastic mold during the injection molding process is obtained by fusing the first temperature result and the second temperature result.

[0047] Furthermore, the injection molding monitoring system fuses the first temperature result and the second temperature result to obtain final temperature information of the target plastic mold during the injection molding process, as specifically shown in steps 401 to 404 .

[0048] By setting up multiple machine vision-based visible light cameras and infrared cameras above and on the sides of the injection molding area, the embodiment of the present invention can fully cover the injection mold from different angles and positions, obtain a large amount of mold surface information, and then analyze a more comprehensive and detailed temperature distribution, and capture local subtle temperature differences. On the other hand, the temperature information of the surface visible light image and infrared thermal image is predicted separately through the temperature recognition model, and then the temperature information of the surface visible light image and infrared thermal image are fused to obtain the final temperature information of the plastic mold during the injection molding process, thereby reducing temperature errors. Therefore, the embodiment of the present invention improves the accuracy and stability of temperature measurement, achieves improved temperature measurement precision, and improves the accuracy of temperature monitoring of the plastic mold during the injection molding process.

[0049] In one embodiment, steps 201 to 204 are described as follows:

[0050] Step 201 : filtering each pixel in the surface visible light image based on the local pixel mean and local pixel standard deviation of each pixel in a local neighborhood of a preset size to obtain a preprocessed image corresponding to the surface visible light image.

[0051] Optionally, first collect the surface visible light image I vis Preprocessing is performed to reduce noise and enhance image contrast. The embodiment of the present invention adopts a local adaptive filtering method for preprocessing. vis For each pixel point (x, y), its local neighborhood N(x, y) is of size m*n.

[0052] Optionally, the injection molding monitoring system calculates the local pixel mean μ(x,y) and the local pixel standard deviation σ(x,y) of the pixels in the local neighborhood N(x,y). The specific formula is as follows:

[0053]

[0054] Among them, I vis (x,y) represents the surface visible light image I vis The pixel value of the pixel at the (x,y) coordinate.

[0055] Furthermore, the injection molding monitoring system filters the pixels according to the local pixel mean μ(x, y) and the local pixel standard deviation σ(x, y) to obtain the preprocessed image I pre , the specific formula is as follows:

[0056]

[0057] Among them, I pre (x,y) represents the preprocessed image I pre The pixel value of the pixel point at the (x, y) coordinate, α represents the preset adjustment coefficient.

[0058] In step 202, based on the local frequency response of each pixel in the preprocessed image at multiple different frequencies and directions, the gradient magnitude and direction of each pixel at different frequencies and directions are determined, and the local frequency response feature vector obtained by combining the gradient magnitude and direction of each pixel is analyzed to obtain the texture features of the target plastic mold.

[0059] Furthermore, in order to extract the texture features of the mold surface, the pre-processed image I pre , at each pixel point (x, y), calculate its local frequency response at multiple different frequencies and directions.

[0060] The embodiment of the present invention uses a group of Gaussian filters G with different frequencies. ω (x,y)(ω represents frequency) for the preprocessed image I pre Perform convolution operation to obtain the response image I of different frequency channels ω , the specific formula is as follows:

[0061] I ω (x,y)=I pre (x,y)*G ω (x,y). Where I ω (x,y) represents the response image I ω The pixel value of the pixel at the (x, y) coordinate, * indicates the convolution operation.

[0062] Furthermore, on each frequency channel ω, the gradient magnitude M in r different directions θ (θ = 1, 2, ..., r) is calculated. ω,θ (x,y) and the gradient direction Φω,θ (x,y), the specific formula is as follows:

[0063]

[0064] in, and represents the partial derivatives of the horizontal and vertical axes in the direction θ.

[0065] Furthermore, the injection molding monitoring system combines the gradient amplitude and direction of each pixel at different frequencies and directions to obtain the local frequency response feature vector LFRD(x,y). The specific formula is as follows:

[0066]

[0067] Among them, ω n Representing different frequencies, the texture characteristics of the mold surface are obtained by statistically analyzing the local frequency response feature vector LFRD(x,y) of all pixel points.

[0068] Step 203 : Based on the color distribution information of each pixel in the preprocessed image in the preset color space, a local color histogram of each pixel is determined, and the local color histogram of each pixel is analyzed to obtain the color characteristics of the target plastic mold.

[0069] Furthermore, for color feature extraction, the embodiment of the present invention adopts a method based on color space transformation and local color distribution. The injection molding monitoring system converts the image from the RGB color space to a new color space. In one embodiment, the color space (X, Y, Z) is.

[0070] X=a1R+b1G+c1B.

[0071] Y=a2R+b2G+c2B.

[0072] Z=a3R+b3G+c3B.

[0073] Among them, a i ,b i ,c i (i=1, 2, 3) represents the preset coefficients used to optimize color differentiation.

[0074] In the new color space, for each pixel (x, y), calculate its local color histogram H(x, y) to describe the color distribution information around the point. The calculation of the local color histogram is based on a local area N centered at (x, y). * (x,y), count the frequency of occurrence of different color components in the area.

[0075] Furthermore, the injection molding monitoring system combines and counts the local color histograms of all pixels to obtain the color characteristics of the mold surface.

[0076] In step 204 , the local brightness second derivative and reflectivity are determined based on the local brightness of each pixel in the preprocessed image, and the glossiness feature vector obtained by combining the local brightness second derivative and reflectivity of each pixel is analyzed to obtain the glossiness feature of the target plastic mold.

[0077] Furthermore, for glossiness feature extraction, the embodiment of the present invention is based on the local brightness change and reflection characteristics of the image. pre , the injection molding monitoring system calculates the local brightness L(x,y) of each pixel (x,y). The specific formula is as follows:

[0078] Among them, R(x,y), G(x,y), B(x,y) represent the preprocessed image I pre The RGB components of the pixel at coordinates (x,y).

[0079] Furthermore, at each pixel point (x, y), the injection molding monitoring system calculates the second-order derivative of its local brightness The specific formula is as follows:

[0080]

[0081] The second-order derivative of local brightness reflects the intensity of brightness changes and has a certain relationship with glossiness. At the same time, considering the reflection characteristics, the reflectivity ρ(x,y) of each pixel (x,y) is calculated. The specific formula is as follows:

[0082] ρ(x,y)=(L(x,y)) / (L max (x,y)).

[0083] Among them, L max (x,y) represents the maximum brightness value in the local area where the pixel point (x,y) is located.

[0084] Furthermore, the injection molding monitoring system converts the local brightness second-order derivative Combined with the reflectivity ρ(x,y), we get the glossiness feature vector GL(x,y). The specific formula is as follows:

[0085]

[0086] Furthermore, the injection molding monitoring system obtains the gloss characteristics of the mold surface by analyzing the gloss feature vectors GL(x,y) of all pixel points.

[0087] The embodiment of the present invention analyzes the surface visible light image features to obtain the texture features, color features and gloss features of the mold surface, which can obtain a large amount of mold surface information, and then analyze a more comprehensive and detailed temperature distribution, which can capture local subtle temperature differences, improve the temperature measurement accuracy, and improve the accuracy of temperature monitoring of plastic molds during the injection molding process.

[0088] In one embodiment, steps 205 to 208 are described as follows:

[0089] In step 205 , the local frequency response eigenvectors are grouped according to different frequencies and directions to obtain multiple sub-eigenvectors, and the multiple sub-eigenvectors are analyzed based on the network layer to obtain the temperature component corresponding to the texture.

[0090] Optionally, for the temperature component corresponding to the texture, the injection molding monitoring system groups the local frequency response eigenvector LFRD(x,y) according to different frequencies and directions to obtain multiple sub-eigenvectors LFRD q (x, y), q = 1, 2, ... Q, where Q is the number of groups.

[0091] Furthermore, for each sub-feature vector LFRD q (x,y), passes through a local feature processing layer, which consists of multiple neurons. The calculation method of the neurons is as follows:

[0092]

[0093] Among them, h q,i represents the output of the i-th neuron after the q-th group of sub-feature vectors pass through the local feature processing layer, σ represents the activation function (such as ReLU function), and w q,i,j Represents the connection weight, LFRD q,j (x,y) represents the jth element of the qth group of sub-feature vectors, b q,i,j represents the bias, J q Represents the dimension of the qth group of sub-feature vectors.

[0094] Furthermore, the outputs of the local feature processing layers of all groups are concatenated to obtain the global feature vector C(x,y), which is passed through a global feature fusion layer. The calculation of this layer is:

[0095]

[0096] Among them, D f represents the output of the global feature fusion layer, β q,i represents the fusion weight, I q Represents the number of neurons in the qth group of local feature processing layer.

[0097] Furthermore, the temperature component T corresponding to the texture is obtained through the output layer texture , the specific formula is as follows:

[0098] T texture =γ·T texture +δ.

[0099] Among them, γ and δ represent the parameters of the output layer.

[0100] Step 206 , mapping the high-dimensional color feature vector in the color feature vector obtained after the local color histogram vectorization processing to a low-dimensional manifold space, and determining the temperature component corresponding to the color based on the low-dimensional manifold space and the mapping relationship between color and temperature.

[0101] Optionally, for the temperature component corresponding to the color, the injection molding monitoring system vectorizes the local color histogram H(x, y) of all pixel points to obtain a color feature vector C(x, y).

[0102] Furthermore, the injection molding monitoring system maps the high-dimensional color feature vector to the low-dimensional manifold space C low (x, y). In the low-dimensional manifold space, the mapping relationship between color and temperature is established through a polynomial regression model, where the polynomial regression model is as follows:

[0103]

[0104] Among them, T color Indicates the temperature component corresponding to the color, K T Indicates the highest degree of the polynomial, D T represents the dimension of the low-dimensional manifold space, are the polynomial coefficients, C low,i Represents the i-th element of the color feature vector in the low-dimensional manifold space.

[0105] Step 207 : constructing a joint probability distribution between the glossiness feature vector and temperature based on the glossiness feature vector, and performing temperature value estimation analysis on the joint probability distribution based on maximum a posteriori estimation to obtain a temperature component corresponding to the glossiness.

[0106] Optionally, for the temperature component corresponding to the glossiness, the injection molding monitoring system can calculate the glossiness characteristic vector Construct the joint probability distribution of glossiness feature vector and temperature. The specific formula is as follows:

[0107]

[0108] Among them, P(T gloss|GL) represents the posterior probability of temperature given the known glossiness feature vector GL(x,y), P(GL|T gloss ) represents the likelihood probability of the gloss feature vector GL(x,y) when the temperature is known, P(T gloss ) represents the prior probability of temperature, and P(GL) represents the marginal probability of the glossiness feature vector.

[0109] Furthermore, the injection molding monitoring system determines the most likely temperature value through maximum a posteriori estimation and obtains the temperature component corresponding to the glossiness. The specific formula is as follows:

[0110]

[0111] Among them, T gloss Indicates the temperature component corresponding to glossiness.

[0112] Step 208 : Fusing the temperature component corresponding to the texture, the temperature component corresponding to the color, and the temperature component corresponding to the glossiness, and outputting the first temperature result.

[0113] Furthermore, the injection molding monitoring system calculates the temperature component T corresponding to the texture. texture , the temperature component T corresponding to the color colr The temperature component T corresponding to glossiness gloss The first temperature result is obtained by fusion. The specific analysis is as follows: The injection molding monitoring system calculates the temperature component T corresponding to the texture texture , the temperature component T corresponding to the color color The temperature component T corresponding to glossiness gloss The probability distribution P texture (t), probability distribution P color (t) and the probability distribution P gloss (t), the calculation process will not be described in detail here.

[0114] Furthermore, the injection molding monitoring system is based on the probability distribution P texture (t), probability distribution P color (t) and the probability distribution P gloss (t) Calculate information entropy E texture , information entropy E color and information entropy E gloss , the specific formula is as follows:

[0115] E texture =-∑ t P texture (t)·log2P texture (t).

[0116] E color =-∑ t Pcolor (t)·log2P color (t).

[0117] E gloss =-∑ t P gloss (t)·log2P gloss (t).

[0118] Furthermore, the injection molding monitoring system has a large impact on the information entropy E texture , information entropy E color and information entropy E gloss Perform normalization processing to obtain the normalization coefficient n texture , normalization coefficient n color and normalization coefficient n gloss , the specific formula is as follows:

[0119]

[0120] Furthermore, the injection molding monitoring system is based on the temperature component T corresponding to the texture. texture , the temperature component T corresponding to the color color The temperature component T corresponding to glossiness gloss , and its corresponding normalization coefficient n texture , normalization coefficient n color and normalization coefficient n gloss The first temperature result is calculated, and the specific formula is as follows:

[0121] T1=n texture ·T texture +n color ·T color +n gloss ·T gloss .

[0122] Wherein, T1 represents the first temperature result.

[0123] The embodiment of the present invention performs temperature recognition on the surface visible light image through a temperature recognition model to obtain the temperature component corresponding to the texture, the temperature component corresponding to the color, and the temperature component corresponding to the glossiness. The temperature component corresponding to the texture, the temperature component corresponding to the color, and the temperature component corresponding to the glossiness are then integrated to obtain the temperature information of the plastic mold during the injection molding process, thereby reducing the temperature error, improving the accuracy and stability of the temperature measurement, and improving the accuracy of the temperature monitoring of the plastic mold during the injection molding process.

[0124] In one embodiment, steps 301 to 303 are described as follows:

[0125] Step 301 : performing image contrast enhancement processing on a plurality of initial infrared thermal images to obtain enhanced images corresponding to the plurality of initial infrared thermal images.

[0126] Optionally, before fusing multiple initial infrared thermal images, each initial infrared thermal image needs to be pre-processed. Where k = 1, 2, ..., N, N is the number of images, represents the kth initial infrared thermal image The grayscale value of the pixel at the image pixel coordinate (x, y). The embodiment of the present invention enhances each initial infrared thermal image based on the local adaptive histogram equalization method. The image contrast of the kth initial infrared thermal image is analyzed as follows: In one embodiment, The local area centered at pixel (c,y) is R k (x,y), whose size is m*m. The kth initial infrared thermal image The local area R k The histogram of (x,y) is H k (u), where u represents the gray level.

[0127] Furthermore, the injection molding monitoring system calculates the kth initial infrared thermal image The local area R k The cumulative distribution function of (x,y) is as follows:

[0128]

[0129] Among them, CDF k (u) represents the cumulative distribution function of the local area in the kth initial infrared thermal image, and L represents the total number of gray levels in the image.

[0130] Furthermore, the injection molding monitoring system uses the cumulative distribution function CDF k (u) For the local area R k The pixels within (x, y) are grayscale transformed to obtain enhanced pixels. The specific formula is as follows:

[0131]

[0132] in, Represents the pixel value after enhancement processing, represents a sequence of enhanced images of multiple initial infrared thermal images, Indicates the floor symbol.

[0133] Step 302 : registering the multiple enhanced images based on their angle information and position information to obtain registered images corresponding to the multiple enhanced images.

[0134] Furthermore, in order to fuse images at different angles and different positions and to register the images, an embodiment of the present invention adopts a registration method based on image feature similarity, and registers multiple enhanced images according to the angle information and position information of the multiple enhanced images to obtain registered images corresponding to the multiple enhanced images, as specifically described in steps 3021 to 3025.

[0135] Step 303 : fusing the multiple registered images based on the local information entropy of each registered image to obtain the fused infrared thermal image.

[0136] Furthermore, for each registered image Calculate the registered image The local information entropy E k (x,y), the specific formula is as follows:

[0137] Among them, p k (x, y, u) represents the probability of gray level u appearing in the local area centered on pixel (x, y).

[0138] Further, calculate the weight ω of each pixel of each image k (x,y), the specific formula is as follows:

[0139]

[0140] Furthermore, according to the weight ω k (x,y) will be the registered image Perform fusion to obtain the fused infrared thermal image I fused , the specific formula is as follows:

[0141]

[0142] Among them, I fused (x,y) represents the fused infrared thermal image I fused The pixel value at the pixel position (x,y).

[0143] The embodiment of the present invention fuses infrared thermal images from different angles and positions, which, on the one hand, reduces the error in image acquisition, and on the other hand, can obtain a large amount of mold surface information, and then analyze a more comprehensive and detailed temperature distribution, which can capture subtle local temperature differences, improve the accuracy and stability of temperature measurement, and improve the accuracy of temperature monitoring of plastic molds during the injection molding process.

[0144] In one embodiment, steps 3021 to 3025 are described as follows:

[0145] Step 3021, based on the angle information and position information of each enhanced image, determine the phase consistency of each enhanced image in different directions, and combine the phase consistency of each enhanced image in different directions to obtain the local phase consistency feature vector of each enhanced image.

[0146] Optionally, for each pixel in the enhanced image of each initial infrared thermal image Injection molding monitoring system counting pixels Phase consistency PC in d (d=1,2,...,D) directions k (x,y,d), the specific formula is as follows:

[0147]

[0148] Where f represents the frequency, represents the phase at frequency f and direction d; W k (x,y,f,d) represents the weight function, which is used to balance the contributions of different frequencies and directions.

[0149] Furthermore, the injection molding monitoring system combines the phase consistency in all directions to obtain pixel The local phase consistency eigenvector LPCF of k (x,y), specifically expressed as follows:

[0150] LPCF k (x,y)=[PC k (x,y,1),PC k (x,y,2),...,PC k (x,y,D)].

[0151] Step 3022: For a first enhanced image of any two enhanced images, perform a transformation operation on the first enhanced image based on the initial registration transformation parameters to obtain a transformed image corresponding to the first enhanced image.

[0152] Furthermore, for any two enhanced images in the embodiment of the present invention, the images and images In the embodiment of the present invention, the image is the first enhanced image, image is the second enhanced image. Optionally, initialize the registration transformation parameter T kl , the transformation includes translation (t x ,t y ), rotation (θz ) and zoom(s z ), then initialize the registration transformation parameters Typically, the translation is set to (0,0), the rotation angle is set to 0, and the scaling factor is set to 1.

[0153] Optionally, the embodiment of the present invention is based on the local phase consistency feature vector LPCF between different images k The similarity of (x, y) determines the registration transformation. Therefore, the injection molding monitoring system initializes the registration transformation parameters T kl Perform a transformation operation on one of the images. Perform the transformation operation to obtain the transformed image

[0154] For translation transformation, the new coordinates (x * ,y * ) and the original coordinates (x, y) are related to x * =x+t x ,y * =y+t y ; For rotation transformation, through the rotation matrix Transform the coordinates; for scaling transformations, x * =x*s z ,y * =y*s z Through the above transformation operation, the image Transform to a new position, orientation, and scale.

[0155] Step 3023: Determine a similarity measure between the transformed image and the second enhanced image based on the local phase consistency feature vector of the transformed image and the local phase consistency feature vector of the second enhanced image.

[0156] Furthermore, the transformed image and another image Injection molding monitoring system calculation image and images The similarity measure S between kl , the specific formula is as follows:

[0157]

[0158] Among them, LPCF k,transformed (x,y) represents the transformed image The local phase consistency eigenvector of .

[0159] Step 3024: Update the initial registration transformation parameters based on the similarity metric to obtain final registration transformation parameters.

[0160] Furthermore, the injection molding monitoring system is based on the similarity metric S kl Update the registration transformation parameters T kl The embodiment of the present invention adopts the gradient descent method, therefore, the similarity measure S is calculated. kl About the registration transformation parameter T kl Gradient Update the registration transformation parameter T in the opposite direction of the gradient kl , the specific formula is as follows:

[0161]

[0162] Among them, θ represents the parameter learning rate, which controls the step size of each parameter update.

[0163] Furthermore, the above steps are executed cyclically to continuously update the registration transformation parameter T kl And calculate the similarity measure S kl , until the similarity measure S kl When the termination condition is met, the current registration transformation parameter T kl is the final registration transformation parameter.

[0164] Step 3025 , registering any two enhanced images based on the final registration transformation parameters to obtain registered images corresponding to the multiple enhanced images.

[0165] Furthermore, the injection molding monitoring system transforms the parameters T kl For images Perform the transformation to obtain the registered image At this point, the registered image With image It has good consistency in space and can be used for subsequent image fusion operations.

[0166] The embodiment of the present invention reduces image acquisition errors by aligning multiple images at different angles and positions, providing accurate image data for subsequent temperature prediction, enabling more accurate prediction of the temperature of the plastic mold during the injection molding process, improving the accuracy and stability of temperature measurement, and improving the accuracy of temperature monitoring of the plastic mold during the injection molding process.

[0167] In one embodiment, steps 401 to 404 are described as follows:

[0168] Step 401 : Process the first temperature result and the second temperature result to respectively determine a first probability distribution and a second probability distribution of the first temperature result and the second temperature result within a preset temperature range.

[0169] Optionally, the injection molding monitoring system processes the first temperature result T1 and the second temperature result T2 separately to construct a first probability distribution of the first temperature result T1 within a preset temperature range, and a second probability distribution of the second temperature result T2 within a preset temperature range. In one embodiment, the temperature range is divided into m equally spaced temperature ranges, so the preset temperature ranges in the embodiment of the present invention are [t1, t2, ..., t m For the first temperature result T1, the number of times n that the first temperature result T1 falls in each preset temperature interval within a period of time or under a certain number of samples is counted. 1i (i=1,2,...m), obtain the first probability distribution of the first temperature result T1 within the preset temperature range Similarly, for the second temperature result T2, the number of times n that the second temperature result T2 falls in each preset temperature range is counted. 2i , obtain the second probability distribution of the second temperature result T2 in the preset temperature range

[0170] Step 402: Processing is performed based on the first probability distribution and the second probability distribution to obtain a first relative entropy and a second relative entropy, and a cross distance between the first probability distribution and the second probability distribution;

[0171] Furthermore, in order to measure the difference in probability distribution between the first temperature result T1 and the second temperature result T2, it is necessary to determine the first probability distribution P1(t i ) and the second probability distribution P2(t i ) difference metric, therefore, the injection molding monitoring system calculates the first probability distribution P1(t i )’s first relative entropy (Kullback-Leibler) and the second probability distribution P2(t i )’s second relative entropy, the specific calculation formula is as follows:

[0172]

[0173] Among them, KL 12 Denotes the first probability distribution P1(t i )'s first relative entropy, KL 21 Denotes the second probability distribution P2(t i )'s second relative entropy.

[0174] Furthermore, according to the first probability distribution P1(t i ) and the second probability distribution P2(t i ) Calculate the first probability distribution P1(t i ) and the second probability distribution P2(t iThe specific formula is as follows:

[0175]

[0176] Wherein, CD represents the first probability distribution P1(t i ) and the second probability distribution P2(t i ) between the intersection distances.

[0177] Step 403: Determine a difference measurement index based on the first relative entropy, the second relative entropy, and the cross distance.

[0178] Furthermore, the injection molding monitoring system is based on the first probability distribution P1(t i )’s first relative entropy, the second probability distribution P2(t i )’s second relative entropy, and the first probability distribution P1(t i ) and the second probability distribution P2(t i ) and calculate the first probability distribution P1(t i ) and the second probability distribution P2(t i ), the difference measurement indicator between them is as follows:

[0179]

[0180] Among them, Dif represents the difference measurement index, and α represents the preset adjustment parameter.

[0181] Step 404 : A final temperature information of the target plastic mold during the injection molding process is obtained by fusing the first probability distribution, the second probability distribution, and the difference metric.

[0182] Furthermore, the injection molding monitoring system integrates the first probability distribution, the second probability distribution, and the difference measurement index to obtain the final temperature information of the target plastic mold during the injection molding process, as specifically described in steps 4041 to 4044 .

[0183] The embodiment of the present invention fuses the temperature information of the surface visible light image and the infrared thermal image to obtain the final temperature information of the plastic mold during the injection molding process, thereby reducing temperature errors, improving the accuracy and stability of temperature measurement, and improving the accuracy of temperature monitoring of the plastic mold during the injection molding process.

[0184] In one embodiment, steps 4041 to 4044 are described as follows:

[0185] Step 4041: Fusing the first probability distribution and the second probability distribution to obtain a fused temperature probability distribution.

[0186] Optionally, in order to minimize the difference between the fused temperature probability distribution and the distribution of the first temperature result T1 and the second temperature result T2, it is necessary to determine an optimal fusion parameter.

[0187] Therefore, the injection molding monitoring system converts the first probability distribution P1(t i ) and the second probability distribution P2(t i ) are fused to obtain the temperature probability distribution after fusion, and the formula is as follows:

[0188] P f (t i )=λ*P1(t i )+(1-λ)*P2(t i ).

[0189] Among them, P f (t i ) represents the temperature probability distribution after fusion, and λ represents the fusion parameter to be adjusted.

[0190] Step 4042: construct a target optimization function based on the first probability distribution, the second probability distribution, and the fused temperature probability distribution.

[0191] Furthermore, the injection molding monitoring system is based on the first probability distribution P1(t i ), the second probability distribution P2(t i ), and the temperature probability distribution P after fusion f (t i ), construct the target optimization function J(λ), where the target optimization function can be expressed as:

[0192]

[0193] Step 4043: Based on the gradient descent algorithm, the difference metric target optimization function is solved to obtain the optimal fusion parameters.

[0194] Furthermore, the injection molding monitoring system solves the minimum value of the target optimization function J(λ) with respect to λ. In the embodiment of the present invention, the minimum value of the target optimization function J(λ) with respect to λ is solved based on the gradient descent algorithm, that is: Get the optimal fusion parameter λ out , where λ k represents the fusion parameter value of the kth iteration, Represents the fusion parameter value λ k The gradient descent value of .

[0195] In step 4044 , the first temperature result and the second temperature result are fused based on the optimal fusion parameters to obtain final temperature information of the target plastic mold during the injection molding process.

[0196] Furthermore, the injection molding monitoring system is based on the optimal fusion parameter λ out The first temperature result T1 and the second temperature result T2 are combined to obtain the final temperature information of the target plastic mold during the injection molding process. The specific formula is as follows: final =λ out *T1+(1-λ out )*T2.

[0197] Among them, T final Indicates the final temperature information of the target plastic mold during the injection molding process.

[0198] The embodiment of the present invention fuses the temperature information of the surface visible light image and the infrared thermal image to obtain the final temperature information of the plastic mold during the injection molding process, thereby reducing temperature errors, improving the accuracy and stability of temperature measurement, and improving the accuracy of temperature monitoring of the plastic mold during the injection molding process.

[0199] See also Figure 3 , Figure 3 This is a diagram of an embodiment of an electronic device provided by an embodiment of the present invention. Figure 3 As shown, an embodiment of the present invention provides an electronic device 300, including a memory 310, a processor 320, and a computer program 311 stored in the memory 310 and executable on the processor 320. When the processor 320 executes the computer program 311, the following steps are implemented:

[0200] A plurality of machine vision cameras installed above and on the side of the injection area of ​​the plastic mold are used to collect a surface visible light image and a plurality of initial infrared thermal images of the target plastic mold;

[0201] Performing feature analysis on the surface visible light image to obtain surface features of the target plastic mold, and inputting the surface features into a first temperature recognition model to obtain a first temperature result output by the first temperature recognition model; the first temperature recognition model is trained based on the sample surface features and their corresponding first temperature result labels;

[0202] fusing the multiple initial infrared thermal images to obtain a fused infrared thermal image, and inputting the fused infrared thermal image into a second temperature recognition model to obtain a second temperature result output by the second temperature recognition model; the second temperature recognition model is trained based on the sample infrared thermal images and their corresponding second temperature result labels;

[0203] Based on the fusion of the first temperature result and the second temperature result, the final temperature information of the target plastic mold during the injection molding process is obtained.

[0204] See also Figure 4 , Figure 4Detailed description of an embodiment of a computer-readable storage medium provided by an embodiment of the present invention. Figure 4 As shown, this embodiment provides a computer-readable storage medium 400 on which a computer program 311 is stored. When the computer program 311 is executed by a processor, the following steps are implemented:

[0205] A plurality of machine vision cameras installed above and on the side of the injection area of ​​the plastic mold are used to collect a surface visible light image and a plurality of initial infrared thermal images of the target plastic mold;

[0206] Performing feature analysis on the surface visible light image to obtain surface features of the target plastic mold, and inputting the surface features into a first temperature recognition model to obtain a first temperature result output by the first temperature recognition model; the first temperature recognition model is trained based on the sample surface features and their corresponding first temperature result labels;

[0207] fusing the multiple initial infrared thermal images to obtain a fused infrared thermal image, and inputting the fused infrared thermal image into a second temperature recognition model to obtain a second temperature result output by the second temperature recognition model; the second temperature recognition model is trained based on the sample infrared thermal images and their corresponding second temperature result labels;

[0208] Based on the fusion of the first temperature result and the second temperature result, the final temperature information of the target plastic mold during the injection molding process is obtained.

[0209] On the other hand, the present invention further provides a computer program product, which includes a computer program. The computer program can be stored on a non-transitory computer-readable storage medium. When the computer program is executed by a processor, the computer can perform the plastic mold injection monitoring method provided by the above methods. The plastic mold injection monitoring method includes:

[0210] A plurality of machine vision cameras installed above and on the side of the injection area of ​​the plastic mold are used to collect a surface visible light image and a plurality of initial infrared thermal images of the target plastic mold;

[0211] Performing feature analysis on the surface visible light image to obtain surface features of the target plastic mold, and inputting the surface features into a first temperature recognition model to obtain a first temperature result output by the first temperature recognition model; the first temperature recognition model is trained based on the sample surface features and their corresponding first temperature result labels;

[0212] fusing the multiple initial infrared thermal images to obtain a fused infrared thermal image, and inputting the fused infrared thermal image into a second temperature recognition model to obtain a second temperature result output by the second temperature recognition model; the second temperature recognition model is trained based on the sample infrared thermal images and their corresponding second temperature result labels;

[0213] Based on the fusion of the first temperature result and the second temperature result, the final temperature information of the target plastic mold during the injection molding process is obtained.

[0214] The device embodiments described above are merely illustrative. The units described as separate components may or may not be physically separate, and the components shown as units may or may not be physical units, i.e., they may be located in one location or distributed across multiple network units. Some or all of the modules may be selected based on actual needs to achieve the objectives of the present embodiment. Persons of ordinary skill in the art will be able to understand and implement the present invention without inventive effort.

[0215] Through the description of the above embodiments, those skilled in the art can clearly understand that each embodiment can be implemented by means of software plus a necessary general hardware platform, or of course, by hardware. Based on this understanding, the essence of the above technical solution or the part that contributes to the existing technology can be embodied in the form of a software product. The computer software product can be stored in a computer-readable storage medium, such as ROM / RAM, a magnetic disk, an optical disk, etc., and includes a number of instructions for enabling a computer device (which can be a personal computer, a server, or a network device, etc.) to execute the methods described in each embodiment or certain parts of the embodiments.

[0216] Finally, it should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention, rather than to limit it. Although the present invention has been described in detail with reference to the aforementioned embodiments, those skilled in the art should understand that they can still modify the technical solutions described in the aforementioned embodiments, or make equivalent replacements for some of the technical features therein. However, these modifications or replacements do not deviate the essence of the corresponding technical solutions from the spirit and scope of the technical solutions of the various embodiments of the present invention.

Claims

1. A plastic mold injection monitoring system, characterized in that: It includes an injection molding monitoring center, multiple machine vision camera modules, a first temperature recognition module, a second temperature recognition module, and an injection molding temperature monitoring module; the multiple machine vision camera modules are installed above and on the side of the injection molding area of ​​the plastic mold, and the multiple machine vision camera modules include a visible light camera module and an infrared camera module; the injection molding monitoring center is respectively connected to the multiple machine vision camera modules, the first temperature recognition module, the second temperature recognition module, and the injection molding temperature monitoring module to manage each module; A plurality of the machine vision camera modules are used to capture a surface visible light image and a plurality of initial infrared thermal images of a target plastic mold; The first temperature recognition module is configured to perform feature analysis on the surface visible light image to obtain surface features of the target plastic mold, and input the surface features into a first temperature recognition model to obtain a first temperature result output by the first temperature recognition model; The second temperature recognition module is configured to fuse the multiple initial infrared thermal images to obtain a fused infrared thermal image, and input the fused infrared thermal image into a second temperature recognition model to obtain a second temperature result output by the second temperature recognition model; The injection molding temperature monitoring module is configured to obtain final temperature information of the target plastic mold during the injection molding process by fusing the first temperature result and the second temperature result; The first temperature recognition model is trained based on the surface features of the sample and its corresponding first temperature result label; the second temperature recognition model is trained based on the infrared thermal image of the sample and its corresponding second temperature result label.

2. A plastic mold injection monitoring method, implemented based on the plastic mold injection monitoring system according to claim 1, characterized in that: The plastic mold injection monitoring method comprises: A plurality of machine vision camera modules installed above and on the side of the injection area of ​​the plastic mold are used to collect a surface visible light image and a plurality of initial infrared thermal images of the target plastic mold; performing feature analysis on the surface visible light image to obtain surface features of the target plastic mold, and inputting the surface features into a first temperature recognition model to obtain a first temperature result output by the first temperature recognition model; the first temperature recognition model is trained based on the sample surface features and their corresponding first temperature result labels; fusing the plurality of initial infrared thermal images to obtain a fused infrared thermal image, and inputting the fused infrared thermal image into a second temperature recognition model to obtain a second temperature result output by the second temperature recognition model; the second temperature recognition model is trained based on sample infrared thermal images and their corresponding second temperature result labels; Based on the fusion of the first temperature result and the second temperature result, final temperature information of the target plastic mold during the injection molding process is obtained.

3. The plastic mold injection monitoring method according to claim 2, characterized in that: The fusing of the first temperature result and the second temperature result to obtain final temperature information of the target plastic mold during the injection molding process includes: Processing the first temperature result and the second temperature result to respectively determine a first probability distribution and a second probability distribution of the first temperature result and the second temperature result within a preset temperature range; Performing processing based on the first probability distribution and the second probability distribution to obtain a first relative entropy and a second relative entropy, respectively, and a cross distance between the first probability distribution and the second probability distribution; determining a difference metric based on the first relative entropy, the second relative entropy, and the cross distance; Based on the fusion of the first probability distribution, the second probability distribution and the difference measurement index, the final temperature information of the target plastic mold during the injection molding process is obtained.

4. The plastic mold injection monitoring method according to claim 3, characterized in that: The fusing of the first probability distribution, the second probability distribution, and the difference metric to obtain final temperature information of the target plastic mold during the injection molding process includes: Fusing the first probability distribution and the second probability distribution to obtain a fused temperature probability distribution; constructing a target optimization function based on the first probability distribution, the second probability distribution, and the fused temperature probability distribution; Solving the target optimization function based on the gradient descent algorithm combined with the difference metric to obtain the optimal fusion parameters; The first temperature result and the second temperature result are fused based on the optimal fusion parameter to obtain final temperature information of the target plastic mold during the injection molding process.

5. The plastic mold injection monitoring method according to claim 2, characterized in that: The surface characteristics of the plastic mold include texture characteristics, color characteristics and gloss characteristics; The performing feature analysis on the surface visible light image to obtain the surface features of the target plastic mold includes: performing filtering processing on each pixel in the surface visible light image based on a local pixel mean and a local pixel standard deviation within a local neighborhood of a preset size for each pixel in the surface visible light image to obtain a preprocessed image corresponding to the surface visible light image; Determining the gradient magnitude and direction of each pixel in the preprocessed image at a plurality of different frequencies and directions based on the local frequency response of each pixel in the preprocessed image, and analyzing the local frequency response feature vector obtained by combining the gradient magnitude and direction of each pixel to obtain the texture feature of the target plastic mold; Determining a local color histogram of each pixel based on color distribution information of each pixel in the preprocessed image in a preset color space, and analyzing the local color histogram of each pixel to obtain a color feature of the target plastic mold; The local brightness second-order derivative and reflectivity of each pixel in the preprocessed image are determined based on the local brightness, and the gloss feature vector obtained by combining the local brightness second-order derivative and reflectivity of each pixel is analyzed to obtain the gloss feature of the target plastic mold.

6. The plastic mold injection monitoring method according to claim 5, characterized in that: Inputting the surface features of the plastic mold into a first temperature recognition model to obtain a first temperature result output by the first temperature recognition model includes: Grouping the local frequency response eigenvectors according to different frequencies and directions to obtain multiple sub-eigenvectors, and analyzing the multiple sub-eigenvectors based on a network layer to obtain temperature components corresponding to the texture; Mapping a high-dimensional color feature vector in the color feature vector obtained after vectorization processing of the local color histogram to a low-dimensional manifold space, and determining a temperature component corresponding to the color based on the low-dimensional manifold space and a mapping relationship between color and temperature; constructing a joint probability distribution between the glossiness feature vector and temperature based on the glossiness feature vector, and performing temperature value estimation analysis on the joint probability distribution based on maximum a posteriori estimation to obtain a temperature component corresponding to the glossiness; The temperature component corresponding to the texture, the temperature component corresponding to the color, and the temperature component corresponding to the glossiness are fused to output the first temperature result.

7. The plastic mold injection monitoring method according to any one of claims 2 to 6, characterized in that: The fusing of the plurality of initial infrared thermal images to obtain a fused infrared thermal image comprises: performing image contrast enhancement processing on the plurality of initial infrared thermal images to obtain enhanced images corresponding to the plurality of initial infrared thermal images; Registering the multiple enhanced images based on the angle information and position information of the multiple enhanced images to obtain registered images corresponding to the multiple enhanced images; The multiple registered images are fused based on the local information entropy of each registered image to obtain the fused infrared thermal image.

8. The plastic mold injection monitoring method according to claim 7, characterized in that: The registering the multiple enhanced images based on the angle information and the position information of the multiple enhanced images to obtain the registered images corresponding to the multiple enhanced images includes: Based on the angle information and position information of each enhanced image, the phase consistency of each enhanced image in different directions is determined, and the phase consistency of each enhanced image in different directions is combined to obtain a local phase consistency feature vector of each enhanced image; For a first enhanced image of any two enhanced images, performing a transformation operation on the first enhanced image based on the initial registration transformation parameters to obtain a transformed image corresponding to the first enhanced image; the transformation operation includes translation, rotation and scaling; determining a similarity measure between the transformed image and the second enhanced image based on a local phase consistency feature vector of the transformed image and a local phase consistency feature vector of a second enhanced image, wherein the second enhanced image is the other image of any two enhanced images; updating the initial registration transformation parameters based on the similarity metric to obtain final registration transformation parameters; Any two enhanced images are registered based on the final registration transformation parameters to obtain registered images corresponding to multiple enhanced images.

9. An electronic device comprising: The memory and the processor are characterized in that a computer software program is stored in the memory, and when the processor reads and executes the computer software program, the plastic mold injection monitoring method according to any one of claims 2 to 8 is implemented.

10. A non-transitory computer-readable storage medium, characterized in that The storage medium stores a computer software program, and when the computer software program is executed by the processor, the plastic mold injection monitoring method according to any one of claims 2 to 8 is implemented.

Images