Image color correction method and system for intelligent image processing

Abstract

The invention provides an image color correction method and system for intelligent image processing, and relates to the technical field of image color correction. The method comprises the following steps: when an unmanned intelligent turntable rotates and scans in an environment containing a metal reflective object, acquiring environmental light wavelength distribution data and polarization angle change data; performing spatial mapping on the two types of data to generate a wavelength and polarization direction dynamic matrix; a programmable polarization modulation layer is integrated, a dynamic matrix is called to drive the modulation layer to switch the transmission polarization direction, and interference of high reflection of the metal surface on the original color is restrained; performing red-green-blue three-channel separation on the image after polarization processing, and calculating a light transmission attenuation coefficient caused by metal reflection based on pixel intensity distribution of each channel; and establishing a reverse correlation model of the light transmission attenuation coefficient and the color compensation intensity, and dynamically generating a light and shade conversion characteristic parameter of each channel pixel by inputting the coefficient. According to the invention, the image color rendition precision and the dynamic adaptive capacity in a metal high-reflection scene are improved.

Description

Technical Field

[0001] This application relates to the field of image color correction technology, and in particular to an image color correction method and system for intelligent image processing. Background Technology

[0002] In the field of industrial quality inspection, when optically inspecting surface defects of metal parts, the strong reflective surfaces on high-speed moving production lines cause frequent overexposure and color distortion problems in the imaging system. It is necessary to suppress the interference of metal reflection and restore the true surface texture features, while adapting to the dynamic changes in production line lighting conditions and the differences in spectral reflectance of different metal materials.

[0003] The current mainstream solution uses a fixed-angle polarizing filter array to cover the image sensor, filters specific reflected light through a preset single polarization direction, and performs brightness compensation for overexposed areas based on multi-frame image fusion technology, combined with offline calibrated metal material reflection parameters to achieve basic color correction.

[0004] Fixed polarization direction cannot respond to dynamically changing ambient light spectrum characteristics, causing the suppression effect of high reflectivity areas to fail as production line lighting fluctuates; the processing delay introduced by multi-frame fusion is difficult to meet the detection requirements of high-speed production lines; offline calibration parameters cannot adapt to the differences in reflectivity of different metal objects, resulting in loss of texture details and an increase in the misjudgment rate. Summary of the Invention

[0005] This application provides an image color correction method and system for intelligent image processing, which solves the problems of insufficient image color restoration accuracy and dynamic adaptability in high-reflectivity metallic scenes in the prior art.

[0006] In a first aspect, this application provides an image color correction method for intelligent image processing, comprising: When the unmanned intelligent turntable performs rotational scanning in an environment containing metallic reflective objects, it simultaneously collects ambient light wavelength distribution data and polarization angle change data corresponding to the turntable rotation angle. The ambient light wavelength distribution data and polarization angle change data are spatially mapped to generate a dynamic correspondence matrix between wavelength and polarization direction that varies with the turntable angle. A programmable polarization modulation layer is integrated in the optical path in front of the image sensor. The dynamic correspondence matrix is ​​called to drive the programmable polarization modulation layer to switch the transmission polarization direction, so as to suppress the interference of the high reflectivity area of ​​the metal surface on the original image color data. The red, green and blue channels of the image after polarization direction switching are separated, and the light transmission attenuation coefficient caused by metal reflection is calculated based on the pixel intensity distribution of each channel after separation. A reverse correlation model is established between the light transmission attenuation coefficient and the color compensation intensity during the channel separation process. By inputting the light transmission attenuation coefficient into the reverse correlation model, pixel brightness and darkness conversion feature parameters of each color channel are dynamically generated. Color restoration under metallic reflective environment is achieved by updating the pixel brightness and darkness conversion feature parameters.

[0007] Optionally, the step of spatially mapping the ambient light wavelength distribution data with the polarization angle change data to generate a dynamic correspondence matrix between wavelength and polarization direction that varies with the turntable angle includes: A spatial position sequence of the turntable rotation angle is set, and ambient light wavelength distribution data and polarization angle change data are synchronously collected at each spatial position in the spatial position sequence. The ambient light wavelength distribution data at the same spatial location are spatially allocated according to the spectral range to form spectral spatial units, and the highest frequency value of the polarization angle change data in the spectral spatial unit is taken as the specified polarization direction. A spatial correspondence is established between the spectral spatial units and the specified polarization direction, generating a dynamic correspondence matrix between wavelength and polarization direction that varies with the turntable angle.

[0008] Optionally, integrating a programmable polarization modulation layer in the optical path in front of the image sensor, and using the dynamic correspondence matrix to drive the programmable polarization modulation layer to switch the transmission polarization direction to suppress interference from highly reflective areas of the metal surface on the original image color data, includes: The dominant spectral interval is extracted from the ambient light wavelength distribution data, and the target polarization angle is obtained by querying the dynamic correspondence matrix based on the dominant spectral interval. A programmable polarization modulation layer is integrated in the optical path in front of the image sensor, and a drive command carrying the target polarization angle is sent to the programmable polarization modulation layer. According to the driving command, the physical arrangement direction of the internal light adjustment plate is adjusted through the programmable polarization modulation layer so that the transmission polarization direction is aligned with the target polarization angle, thereby suppressing the interference of the high reflectivity area of ​​the metal surface on the original image color data.

[0009] Optionally, establishing an inverse correlation model between the light transmittance attenuation coefficient and the color compensation intensity during the channel separation process, and dynamically generating pixel brightness transition feature parameters for each color channel by inputting the light transmittance attenuation coefficient into the inverse correlation model, and achieving color restoration under metallic reflective environments by updating the pixel brightness transition feature parameters, includes: Extract the proportion of high-value regions and low-value regions of pixel intensity during the channel separation process, and calculate the difference between the proportion of high-value regions and the proportion of low-value regions as the basic light transmission reduction amount; Multiply the basic light transmission reduction by the metal reflection influence factor to output the light transmission attenuation coefficient, and construct an inverse correlation model between the light transmission attenuation coefficient and the color compensation intensity during the channel separation process. The light transmission attenuation coefficient is input into the inverse correlation model, and the color compensation intensity value is output. The pixel brightness and darkness conversion feature parameters of each color channel are generated based on the color compensation intensity value. Color restoration under metallic reflective environment is achieved by updating the pixel brightness and darkness conversion feature parameters.

[0010] Optionally, the step of performing red, green, and blue channel separation on the image after polarization direction switching, and calculating the channel transmittance attenuation coefficient caused by metal reflection based on the pixel intensity distribution of each channel after separation, includes: Color component decomposition is performed on the image after polarization direction switching, and the red component image, green component image and blue component image are extracted respectively. The proportion of the number of pixels with brightness values ​​exceeding the set brightness threshold in each color component image is counted as the proportion of bright pixels. The proportion of dark pixels is calculated as the percentage of pixels in an image with the same color component whose brightness value is lower than a set dark threshold. The numerical difference between the proportion of bright pixels and the proportion of dark pixels is calculated as the base value of channel light transmission loss. The base value of channel light transmission loss is multiplied by the metal reflection correction factor to output the channel light transmission attenuation coefficient caused by metal reflection.

[0011] Optionally, the step of extracting the dominant spectral interval from the ambient light wavelength distribution data and obtaining the target polarization angle by querying the dynamic correspondence matrix based on the dominant spectral interval includes: The ambient light wavelength distribution data is divided into multiple spectral segments according to fixed wavelength intervals, and the sum of the energy values ​​corresponding to all wavelength points in each spectral segment is calculated as the energy concentration degree. The spectral segment containing the largest energy concentration value is selected as the dominant spectral segment; Search the dynamic correspondence matrix for record entries that perfectly match the wavelength range of the dominant spectral region, and extract the corresponding polarization direction record from the record entries as the target polarization angle.

[0012] Optionally, the step of spatially allocating the ambient light wavelength distribution data according to the spectral range to form spectral spatial units at the same spatial location, and using the highest frequency value of the polarization angle change data in the spectral spatial unit as the specified polarization direction, includes: In the same spatial location, the ambient light wavelength distribution data is divided into spectral spatial units according to the progressive relationship of wavelength values ​​corresponding to the spectral range from small to large. Within the wavelength range covered by the spectral spatial unit, statistically analyze the corresponding polarization angle variation data and record the total number of occurrences of the polarization angle variation data within the wavelength range; The total number of occurrences is compared, and the polarization angle change data with the highest total number of occurrences is selected as the specified polarization direction of the spectral spatial unit.

[0013] Secondly, this application provides an image color correction system for intelligent image processing, comprising: The acquisition module is used to simultaneously acquire ambient light wavelength distribution data and polarization angle change data corresponding to the rotation angle of the turntable when the unmanned intelligent turntable performs rotational scanning in an environment containing metallic reflective objects. The mapping module is used to spatially map the ambient light wavelength distribution data with the polarization angle change data to generate a dynamic correspondence matrix between wavelength and polarization direction that varies with the turntable angle. An integrated module is used to integrate a programmable polarization modulation layer in the optical path in front of the image sensor, and to call the dynamic correspondence matrix to drive the programmable polarization modulation layer to switch the transmission polarization direction in order to suppress the interference of the high reflectivity area of ​​the metal surface on the original image color data. The calculation module is used to separate the red, green and blue channels of the image after polarization direction switching, and calculate the channel light transmission attenuation coefficient caused by metal reflection based on the pixel intensity distribution of each channel after separation. The generation module is used to establish an inverse correlation model between the light transmission attenuation coefficient and the color compensation intensity during the channel separation process. By inputting the light transmission attenuation coefficient into the inverse correlation model, the pixel brightness and darkness conversion feature parameters of each color channel are dynamically generated. By updating the pixel brightness and darkness conversion feature parameters, color restoration under metallic reflective environment is achieved.

[0014] Thirdly, this application provides a computing device, including a processing component and a storage component; the storage component stores one or more computer instructions; the one or more computer instructions are to be invoked and executed by the processing component to implement an image color correction method for intelligent image processing as described in the first aspect above.

[0015] Fourthly, this application provides a computer storage medium storing a computer program, which, when executed by a computer, implements an image color correction method for intelligent image processing as described in the first aspect.

[0016] In this embodiment, when an unmanned intelligent turntable performs rotational scanning in an environment containing reflective metallic objects, ambient light wavelength distribution data and polarization angle change data corresponding to the turntable rotation angle are simultaneously collected. The ambient light wavelength distribution data and polarization angle change data are spatially mapped to generate a dynamic correspondence matrix between wavelength and polarization direction that changes with the turntable angle. A programmable polarization modulation layer is integrated in the optical path in front of the image sensor. The dynamic correspondence matrix is ​​used to drive the programmable polarization modulation layer to switch the transmission polarization direction to suppress the interference of highly reflective areas on the original image color data. The image after polarization direction switching is separated into red, green, and blue channels. The channel transmittance attenuation coefficient caused by metal reflection is calculated based on the pixel intensity distribution of each channel after separation. An inverse correlation model is established between the transmittance attenuation coefficient and the color compensation intensity during the channel separation process. By inputting the transmittance attenuation coefficient into the inverse correlation model, pixel brightness conversion feature parameters of each color channel are dynamically generated. Color restoration in a reflective metallic environment is achieved by updating the pixel brightness conversion feature parameters.

[0017] The technical solution of this application has the following beneficial effects: Synchronously acquiring turntable rotation angle data enables dynamic capture of multi-angle spectral and polarization characteristics under metallic reflective environments, providing fundamental physical quantity support for suppressing ambient light interference. A quantitative mapping rule is established for reflection angle, spectral characteristics, and optimal polarization suppression, addressing the problem that traditional fixed polarization strategies cannot adapt to dynamic lighting. Hardware-level polarization control precisely suppresses optical interference in highly reflective areas of the metal surface, preserving the integrity of the original color data. Quantifying the differentiated optical attenuation of RGB channels due to metal reflection provides precise input parameters for color distortion compensation. Pixel-level dynamic color compensation is achieved, eliminating channel imbalance distortion caused by metal reflection and restoring true surface colors.

[0018] Furthermore, a spatial position sequence of the turntable rotation angle is defined, and ambient light wavelength distribution data and polarization angle change data are simultaneously collected at each spatial position in the sequence. Spectral data at the same position are spatially allocated according to range to form spectral spatial units, with the highest frequency value of the polarization angle specifying the polarization direction. A spatial correspondence is established between the spectral spatial units and the specified polarization direction, generating a dynamic correspondence matrix that changes with the angle. A high-precision spectral and polarization correlation library is constructed. Through spatial position binding and frequency statistics algorithms, discrete ambient light data is transformed into queryable polarization control rules, significantly improving the decision-making efficiency and accuracy of the optimal polarization direction in metallic reflection scenarios.

[0019] These or other aspects of this application will become more apparent in the following description of the embodiments. Attached Figure Description

[0020] To more clearly illustrate the technical solutions in the embodiments of this application or the prior art, the drawings used in the description of the embodiments or the prior art will be briefly introduced below. Obviously, the drawings described below are some embodiments of this application. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0021] Figure 1 A flowchart of an image color correction method for intelligent image processing provided in this application is shown; Figure 2 A flowchart of an image color correction method for intelligent image processing provided in this application is shown; Figure 3 This application provides a schematic diagram of the structure of an image color correction system for intelligent image processing. Figure 4 A schematic diagram of the structure of a computing device provided in this application is shown. Detailed Implementation

[0022] To enable those skilled in the art to better understand the present application, the technical solutions in the embodiments of the present application will be clearly and completely described below with reference to the accompanying drawings.

[0023] In some of the processes described in the specification, claims, and accompanying drawings of this application, multiple operations appearing in a specific order are included. However, it should be clearly understood that these operations may not be executed in the order they appear herein, or may be executed in parallel. The operation numbers, such as 101, 102, etc., are merely used to distinguish different operations and do not themselves represent any execution order. Furthermore, these processes may include more or fewer operations, and these operations may be executed sequentially or in parallel. It should be noted that the descriptions such as "first," "second," etc., in this document are used to distinguish different messages, devices, modules, etc., and do not represent a chronological order, nor do they limit "first" and "second" to different types.

[0024] In optical inspection of metal parts, the static suppression mechanism of fixed-angle polarization filters is ill-suited to dynamic fluctuations in production line lighting, causing the suppression effect in high-reflectivity areas to fail as ambient light changes. Multi-frame fusion compensation methods introduce processing delays exceeding 100 milliseconds, failing to meet the demands of high-speed production line inspection. Furthermore, calibration parameters relying on offline calibration lack material adaptability, leading to texture detail loss and increased false positive rates in scenarios involving mixed lines of dissimilar metals such as aluminum and stainless steel. These three shortcomings collectively point to the fundamental limitations of traditional solutions in terms of dynamic adaptability, reliability, and universality.

[0025] To address the aforementioned shortcomings, this invention proposes a metal surface color correction method based on ambient light parameter perception. Its core lies in constructing a synergistic system of spectral and polarization dynamic response chains and channel-level negative feedback compensation mechanisms. Multi-angle spectral-polarization coupled data is simultaneously captured through rotating scanning, generating environmentally responsive polarization control rules to drive the electronically controlled polarization unit to switch to the optimal suppression direction. Combined with channel transmittance attenuation quantization and a reverse compensation model within a single frame image, millisecond-level closed-loop elimination of metal reflection interference is achieved. This method overcomes the static limitations of fixed polarization strategies, replacing multi-frame fusion with environmentally driven polarization control, reducing processing latency. Through a material-insensitive dynamic generation mechanism of attenuation coefficients, it eliminates dependence on offline calibration, achieving high color reproduction accuracy in tests on metal surfaces such as aluminum alloys and galvanized steel, fundamentally solving the problems of lack of specificity and universality in high-reflectivity scenarios.

[0026] The technical solutions of the embodiments of this application will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of this application, and not all embodiments. Based on the embodiments of this application, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of this application.

[0027] Figure 1 This application provides a flowchart of an image color correction method for intelligent image processing, as shown in the embodiments of this application. Figure 1 As shown, the method includes: 101. When the unmanned intelligent turntable performs rotational scanning in an environment containing metallic reflective objects, it simultaneously collects ambient light wavelength distribution data and polarization angle change data corresponding to the turntable rotation angle. In the above scheme, the unmanned intelligent turntable refers to an automatically rotating shooting platform. The environment containing reflective metallic objects specifically refers to scenes with highly reflective metals such as stainless steel and aluminum alloys. Rotational scanning is the action of the turntable rotating gradually at preset angular intervals. Ambient light wavelength distribution data records the intensity ratio of different colors of light. Polarization angle change data describes the change in the direction of light vibration.

[0028] In this embodiment, firstly, an unmanned intelligent turntable is controlled to rotate gradually at fixed angular intervals in a metallic reflective environment. Whenever the turntable rotates to a new angular position, two types of sensors are immediately triggered to work synchronously: a spectral analysis module records the color composition of the ambient light at the current angle; and a polarization detection module captures the offset of the light vibration direction, such as 0 degrees representing vertical vibration and 90 degrees representing horizontal vibration. Secondly, during the turntable rotation, the reflective properties of the metal surface dynamically change with the angle. Finally, each rotation angle position, the corresponding ambient light color composition data, and the light vibration direction data are bound and stored in a data set in a specific storage format. For example, the storage format is: angle 30 degrees - red light intensity value 0.7 / green light 0.2 / blue light 0.1 - polarization angle 60 degrees; forming a data set: {angle 1: color data 1 + polarization data 1, angle 2: color data 2 + polarization data 2, ...}.

[0029] 102. Spatial mapping is performed between the ambient light wavelength distribution data and the polarization angle change data to generate a dynamic correspondence matrix between wavelength and polarization direction that varies with the turntable angle; Optionally, such as Figure 2 As shown, step 102 may specifically include the following steps: 1021. Set the spatial position sequence of the turntable rotation angle, and synchronously collect ambient light wavelength distribution data and polarization angle change data at each spatial position in the spatial position sequence; 1022. At the same spatial location, the ambient light wavelength distribution data is spatially allocated according to the spectral range to form spectral spatial units, and the highest frequency value of the polarization angle change data in the spectral spatial unit is taken as the specified polarization direction. Specifically, step 1022 may include the following process: At the same spatial location, for the ambient light wavelength distribution data, according to the progressive relationship of wavelength values ​​corresponding to the spectral range from small to large, the ambient light wavelength distribution data is divided into spectral spatial units; within the wavelength range covered by the spectral spatial unit, the corresponding polarization angle change data is statistically analyzed, and the total number of occurrences of the polarization angle change data within the wavelength range is recorded; the total number of occurrences is compared, and the polarization angle change data with the highest total number of occurrences is selected as the specified polarization direction of the spectral spatial unit.

[0030] 1023. Establish a spatial correspondence between the spectral spatial unit and the specified polarization direction to generate a dynamic correspondence matrix between wavelength and polarization direction that varies with the turntable angle.

[0031] In the above scheme, the spatial position sequence refers to the set of position points that stop at preset angular intervals during the rotation of the turntable. Ambient light wavelength distribution data is a set of values ​​recording the intensity ratios of different wavelengths of light, specifically including the intensity values ​​of red, green, and blue light within the visible spectrum. Polarization angle change data is a set of values ​​describing the angular changes in the direction of light vibration, obtained through multiple sampling to obtain an angle value sequence. A spectral spatial unit is an independent analysis segment divided into specific wavelength ranges within a continuous spectrum, typically divided into red, green, and blue light units. The highest frequency value refers to the value that appears most frequently in the data set; when frequencies are the same, the average value is taken as the judgment result. The specified polarization direction is the highest value in the frequency statistics of polarization angles within the spectral spatial unit, representing the optimal suppression direction for that spectral unit. The dynamic correspondence matrix is ​​a rule table recording the mapping relationship between the turntable angle, the spectral spatial unit, and the specified polarization direction, using a storage format of "angle: {spectral unit → direction}".

[0032] In this embodiment, firstly, through step 1021, the turntable control system presets a rotation angle sequence. At each angle point, it triggers simultaneous acquisition by two sensors: a spectral sensor scans the visible light band and outputs red / green / blue light intensity values; a polarization sensor performs multiple samples, recording the angle value of the light vibration direction each time. The acquisition process ensures that the angle position, spectral data, and polarization data are synchronized in time. For example, when the turntable stops at 45°, the spectral sensor records a red light intensity of 58 units, a green light intensity of 33 units, and a blue light intensity of 9 units; the polarization sensor samples 10 times to obtain data [30°, 45°, 30°, 30°, 60°, 30°, 45°, 30°, 30°, 30°].

[0033] Then, step 1022 divides the spectral data at the current angle into units: the continuous wavelength range is divided into red light units, green light units, and blue light units. Within each unit, polarization data is statistically analyzed: all polarization sample values ​​corresponding to that unit are extracted; the frequency of each angle value is calculated; the angle with the highest frequency is selected as the designated direction, and if the frequencies are the same, the average value is taken. For example: the polarization data corresponding to the red light unit at the 45° position is [30°, 30°, 30°, 30°, 30°, 30°] (30° appears 6 times in 10 samples), then the designated direction is 30°.

[0034] Finally, step 1023 constructs the matrix entries: creating angle identifiers; binding the relationship between all spectral units at that angle and the specified direction; storing it as "angle: {red unit → direction value, green unit → direction value, blue unit → direction value}". The process is repeated as the turntable rotates to the new angle, ultimately forming a complete dynamic correspondence matrix. For example, the matrix entries for the 45° position are: red unit → 30°, green unit → 45°, blue unit → 60°.

[0035] In practical applications, through step 1021 on the aluminum alloy wheel hub detection line, the turntable rotates in sequence [0°, 45°]. After locking the 0° position: the spectral sensor outputs red light intensity of 70 units / green light of 20 units / blue light of 10 units; the polarization sensor samples data 10 times: [25°, 40°, 25°, 25°, 40°, 25°, 40°, 25°, 25°, 40°].

[0036] Step 1022 processes the 0° position data and divides the spectral units: the red light unit (600-780nm) corresponds to polarization data [25°, 25°, 25°, 25°, 25°, 25°], with 25° appearing 6 times; statistical results: 25° has the highest frequency → red light unit, specified direction 25°; green light unit polarization data [40°, 40°, 40°, 40°] → specified direction 40°.

[0037] Step 1023 generates the matrix's first entry: "0° position: Red light unit → 25°, Green light unit → 40°, Blue light unit → 55°". After the entire process, the turntable rotates to the 45° position. Spectral data: Red light 25 units / Green light 50 units / Blue light 25 units; Polarization data: [60°, 75°, 60°, 75°, 60°, 75°, 60°, 75°, 60°, 75°]; Red light unit polarization data [60°, 60°, 60°, 60°, 60°] → specified direction 60°; Update the matrix: "45° position: Red light unit → 60°, Green light unit → 75°, Blue light unit → 45°". The final matrix contains two core mapping rules.

[0038] The overall scheme of step 102 above establishes a precise matching relationship between the reflectivity of the metal surface and the polarization suppression strategy through angle serialization acquisition and spectral unit frequency analysis, solves the problem of polarization control failure caused by angle changes in rotation detection, and significantly improves the image acquisition stability of highly reflective metal surfaces.

[0039] 103. Integrate a programmable polarization modulation layer in the optical path in front of the image sensor, and call the dynamic correspondence matrix to drive the programmable polarization modulation layer to switch the transmission polarization direction, so as to suppress the interference of the high reflectivity area of ​​the metal surface on the original image color data. Optionally, step 103 may specifically include the following steps: 1031. Extract the dominant spectral interval from the ambient light wavelength distribution data, and obtain the target polarization angle by querying the dynamic correspondence matrix based on the dominant spectral interval; Step 1031 may specifically include the following process: dividing the ambient light wavelength distribution data into multiple spectral segments according to fixed wavelength intervals, calculating the sum of the energy values ​​corresponding to all wavelength points in each spectral segment as the energy concentration degree; selecting the spectral segment containing the largest energy concentration degree as the dominant spectral segment; searching in the dynamic correspondence matrix for record entries that completely match the wavelength range of the dominant spectral segment, and extracting the corresponding polarization direction record from the record entries as the target polarization angle.

[0040] 1032. Integrate a programmable polarization modulation layer in the optical path in front of the image sensor, and send a driving command carrying the target polarization angle to the programmable polarization modulation layer; 1033. According to the driving command, the physical arrangement direction of the internal light adjustment piece is adjusted through the programmable polarization modulation layer so that the transmission polarization direction is aligned with the target polarization angle, so as to suppress the interference of the high reflectivity area of ​​the metal surface on the original image color data.

[0041] In the above scheme, the dominant spectral range is the color range with the strongest energy in ambient light, and its maximum value is determined by calculating the product of the light intensity of each color segment and the wavelength range. The target polarization angle is a specific angle value obtained from the dynamic correspondence matrix, representing the optimal filtering direction under the current environment. The programmable polarization modulation layer is a liquid crystal device installed in front of the lens, which changes its internal structure by receiving electrical signals. The drive command is a control command that converts the target angle into an electrical signal, usually transmitted in digital encoding form. The light adjustment plate is an array of liquid crystal cells inside the polarization modulation layer, and its arrangement direction determines the vibration angle of the light when it passes through. The physical arrangement direction refers to the actual spatial orientation of the liquid crystal cells. The transmitted polarization direction is the vibration angle of the light after passing through the modulation layer, which is consistent with the liquid crystal arrangement direction.

[0042] In this embodiment, firstly, step 1031 obtains the current ambient light color composition data, dividing the spectrum into three color segments: red light (600-780nm), green light (500-600nm), and blue light (380-500nm). The energy values ​​of each segment are calculated: red light energy = red light intensity × 180, green light energy = green light intensity × 100, and blue light energy = blue light intensity × 120. The three energy values ​​are compared, and the color segment with the largest value is selected as the dominant spectral interval. Using the current turntable angle and the dominant spectral segment as an index, the specified direction value stored in the dynamic correspondence matrix is ​​queried; this value is the target polarization angle. For example, if the turntable is stopped at 45°, the ambient light red light intensity is 58 units, the energy is 58 × 180 = 10440, the green light intensity is 33 units / 3300, and the blue light intensity is 9 units / 1080. Red light has the highest energy → the dominant spectral band is red light → look up the corresponding value of the red light unit at position 45° in the matrix → obtain the target angle of 30°.

[0043] Subsequently, in step 1032, a programmable polarization modulation layer is physically installed in front of the optical path of the image sensor. This device includes a liquid crystal cell array and a data interface. The target polarization angle value is converted into an 8-bit binary encoded instruction, for example, 30° is converted into "00011110". The encoded instruction is transmitted to the control port of the modulation layer via a data line. For example: target angle 30° → binary encoding "00011110" → sent to the modulation layer controller via a serial communication interface.

[0044] Finally, in step 1033, the modulation layer controller parses the received binary command and converts it into the corresponding voltage value. A specific voltage is applied to the liquid crystal cell array, such as 3.0V for 30°. Under the influence of the electric field, the liquid crystal molecules physically rotate, gradually turning from the current direction to the target angle. When the liquid crystal cell arrangement direction is consistent with the target angle, the vibration direction of the transmitted light is synchronously aligned, thus suppressing metallic reflection. For example: initial liquid crystal direction 0° → receiving a 30° command → applying a 3.0V voltage → liquid crystal molecules rotate 30° → the vibration direction of the transmitted light changes to 30°.

[0045] In practical applications, assuming the turntable is stopped at 60° on the engine cylinder head inspection line, step 1031 yields the following ambient light data: red light intensity 25 units / energy 4500, green light 50 units / energy 5000, blue light 25 units / energy 3000; green light has the highest energy → the dominant spectral segment is the green light segment; querying the dynamic matrix reveals that the green light unit at the 60° position corresponds to a target angle of 45°.

[0046] In step 1032, the target angle 45° is converted into binary "00101101"; the command is sent to the polarization modulation layer via the data line; the modulation layer is installed 5mm in front of the lens of a 20-megapixel industrial camera.

[0047] In step 1033, the current orientation of the liquid crystal cell is 30°; after the controller interprets the instruction, it applies a voltage of 3.8V; the liquid crystal molecules rotate from 15° to 45°; the vibration direction of the transmitted light is synchronously adjusted to 45°.

[0048] After the entire process, the turntable rotates to a 120° position to collect ambient light: the blue light segment has the highest energy, with an intensity of 68×120=8160; the matrix is ​​queried to obtain the target angle of 80°, and "01010000" / 80° encoding is sent to the modulation layer. The liquid crystal unit rotates from 60° to 80°, and the brightness of the metallic highlight area decreases by 50%.

[0049] The overall solution in step 103 above, through the linkage of ambient light dominant feature recognition and hardware control, dynamically optimizes the polarization filtering direction, effectively eliminates image overexposure and color distortion caused by reflection from metal surfaces, and significantly improves image quality stability in high-reflectivity scenes.

[0050] 104. Perform red, green and blue three-channel separation on the image after polarization direction switching, and calculate the channel transmittance attenuation coefficient caused by metal reflection based on the pixel intensity distribution of each channel after separation. Optionally, step 104 may specifically include the following steps: 1041. Perform color component decomposition on the image after polarization direction switching, and extract the red component image, green component image and blue component image respectively. Calculate the proportion of the number of pixels in each color component image whose brightness value exceeds the set brightness threshold to the total number of pixels as the proportion of bright pixels. 1042. The proportion of dark pixels is the percentage of pixels whose brightness or darkness value is lower than a set darkness threshold in the total number of pixels in the same color component image. 1042. Calculate the numerical difference between the proportion of bright pixels and the proportion of dark pixels as the basic value of channel light transmission loss, multiply the basic value of channel light transmission loss by the metal reflection correction factor, and output the channel light transmission attenuation coefficient caused by metal reflection.

[0051] In the above scheme, the image after polarization direction switching refers to the original acquired image after being filtered by the programmable polarization modulation layer. Red-green-blue three-channel separation is the process of decomposing a color image into three monochrome images: pure red, pure green, and pure blue. Pixel brightness value is the brightness value of each pixel in the monochrome image, ranging from 0 / complete black to 255 / brightest. Highlight threshold is the brightness limit value for determining pixel overexposure. Darkness threshold is the brightness limit value for determining pixel underexposure. Highlight pixel percentage is the percentage of pixels exceeding the highlight threshold out of the total number of pixels in that channel. Darkness pixel percentage is the percentage of pixels below the darkness threshold. The channel transmittance loss base value is the arithmetic difference between the highlight percentage and the darkness percentage. The metal reflection correction factor is an optical adjustment coefficient preset according to the metal type. The channel transmittance attenuation coefficient is the final quantified value of the channel distortion.

[0052] In this embodiment, the pixel image after polarization processing is first obtained in step 1041, and the following operations are performed. First, color separation is performed, extracting the red component: traversing each pixel, only retaining the R value in the RGB to generate a pure red image; the green component: only retaining the G value to generate a pure green image; the blue component: only retaining the B value to generate a pure blue image. Highlight statistics: a highlight threshold is set, all pixels in the pure red image are scanned, the number of pixels with brightness greater than the highlight threshold is counted, and the percentage is calculated as: number of overly bright pixels / number of pixels in the image. For example: 1.12 million overly bright pixels are detected in the pure red image → percentage = 1.12 million / number of pixels in the image 16 million = 7%.

[0053] Then, step 1042 is performed on the same pure red image, setting a dark threshold, scanning all pixels, and counting the number of pixels with brightness less than the dark threshold; the percentage is calculated as: number of excessively dark pixels / number of pixels in the image. For example: if 2.4 million excessively dark pixels are detected, the percentage is 2.4 million / 16 million = 15%.

[0054] Finally, step 1043 calculates the base loss value as the difference between the highlight ratio and the shadow ratio. A negative difference indicates light transmission loss. For example, a highlight ratio of 7% minus a shadow ratio of 15% equals -8%. The metal material coefficient is then obtained, and the light transmission attenuation coefficient is calculated as the absolute value of the difference between the highlight and shadow ratios multiplied by the metal material coefficient. For example, a base value of -8% multiplied by a coefficient of 1.2 results in an output of 9.6%.

[0055] In practical applications, the polarization-processed image of the galvanized steel sheet inspection station is obtained through step 1041, and color component decomposition is performed. A red component image is established: the R value in the RGB values ​​of each pixel is extracted to generate a pure red grayscale image, retaining only red information. For example, the original value of a pixel (R120, G80, B60) becomes (120, 0, 0) after processing. A highlight threshold of 215 is set, and the entire red component image is scanned, comparing the brightness value pixel by pixel. When the pixel brightness > 215, it is counted as a highlight pixel. The cumulative number of highlight pixels is 1.86 million. The proportion of highlight pixels is calculated as: 1.86 million / 20 million = 9.3%. Simultaneously, the green component is processed: threshold 210, detecting 1.54 million highlight pixels → proportion 7.7%; the blue component: threshold 205, detecting 2.21 million → proportion 11.05%.

[0056] Step 1042 performs shadow / darkness statistics on the same red component image: set the shadow / darkness threshold to 45, rescan all pixels of the image: when the pixel brightness value is <45, it is counted in the shadow / darkness pixel counter, and the cumulative number of shadow / darkness pixels is 3.18 million; calculate the percentage of shadow / darkness pixels: 3.18 million / 20 million = 15.9%; green component: threshold 40, detect 4.02 million shadow / darkness pixels → percentage 20.1%; blue component: threshold 35, detect 2.87 million → percentage 14.35%.

[0057] Calculate the base value of light transmission loss for the red channel in step 1043: Highlight ratio 9.3% - Dark ratio 15.9% = -6.6%; Query the metal type database to find the correction factor for galvanized steel sheet as 1.05; Calculate the light transmission attenuation coefficient: Absolute value |-6.6%| × 1.05 = 6.93%; Green channel: 7.7% - 20.1% = -12.4% → Coefficient = 12.4% × 1.05 = 13.02%; Blue channel: 11.05% - 14.35% = -3.3% → Coefficient = 3.3% × 1.05 = 3.465%.

[0058] The overall solution in step 104 above provides key input parameters for color reproduction by accurately quantifying the abnormal brightness distribution of each color channel and dynamically calculating the light transmission loss value in combination with the characteristics of metal materials, effectively solving the problem of color distortion caused by metal reflection.

[0059] 105. Establish an inverse correlation model between the light transmission attenuation coefficient and the color compensation intensity during the channel separation process. By inputting the light transmission attenuation coefficient into the inverse correlation model, dynamically generate pixel brightness and darkness conversion feature parameters for each color channel. By updating the pixel brightness and darkness conversion feature parameters, achieve color restoration under metallic reflective environment.

[0060] Optionally, step 102 may specifically include the following steps: 1051. Extract the proportion of high-value regions and low-value regions of pixel intensity during the channel separation process, and calculate the difference between the proportion of high-value regions and the proportion of low-value regions as the basic light transmission reduction amount; 1052. Multiply the basic light transmission reduction by the metal reflection influence factor to output the light transmission attenuation coefficient, and construct an inverse correlation model between the light transmission attenuation coefficient and the color compensation intensity during the channel separation process. 1053. Input the light transmission attenuation coefficient into the inverse correlation model, output the color compensation intensity value, generate the pixel brightness and darkness conversion feature parameters of each color channel according to the color compensation intensity value, and realize color restoration under metallic reflective environment by updating the pixel brightness and darkness conversion feature parameters.

[0061] In the above scheme, the light transmission attenuation coefficient is a value that quantifies the channel distortion caused by metallic reflection. Color compensation intensity is the strength value used to adjust the brightness of colors. The inverse correlation model is the mathematical relationship where the color compensation intensity decreases as the light transmission attenuation coefficient increases. Pixel brightness conversion feature parameters are the core variables controlling the image brightness curve.

[0062] In this embodiment, step 1051 first performs a brightness distribution analysis on the monochrome image of each color channel: setting a highlight threshold (typically 200-220) and a shadow threshold (typically 30-50); scanning all pixels in the entire monochrome image and counting the number of pixels whose brightness exceeds the highlight threshold; calculating the percentage of highlight pixels to the total number of pixels. The number of pixels whose brightness is below the shadow threshold is also counted, and the percentage of shadow pixels to the total number of pixels is calculated; the ratio of highlight pixels to shadow pixels is subtracted from the ratio of shadow pixels to obtain the basic light transmission reduction. A negative value indicates light transmission loss, while a positive value indicates excessive reflection. For example, in the red channel image, a highlight pixel ratio of 12% minus a shadow pixel ratio of 18% equals -6%.

[0063] Then, step 1052 is used to perform metal reflection characteristic compensation calculation: Based on the metal material type, the preset reflection influence factor is queried: 0.85 for stainless steel and 1.15 for aluminum alloy; the absolute value of the basic light transmission reduction is taken, and multiplied by the metal reflection influence factor to obtain the light transmission attenuation coefficient; an inverse mathematical relationship is constructed: compensation intensity = reference value / (1 + magnification factor × light transmission attenuation coefficient), where the reference value and magnification factor are preset constants. For example: the absolute value of the basic reduction of -6% is 6%, multiplied by the stainless steel factor 0.85 to get 5.1%; setting the reference value to 200 and the magnification factor to 8, then the compensation intensity = 200 / (1 + 8 × 0.051) = 142.

[0064] Finally, dynamic color restoration is achieved through step 1053: the light transmission attenuation coefficient is input into the inverse relational model to calculate the output color compensation intensity value. The compensation intensity value is divided by 100 to obtain the pixel brightness conversion feature parameter (gamma value). The new gamma value is transmitted to the image processing engine, which immediately applies the new parameter to process the current frame image. This process is completed during the turntable rotation interval. For example, a compensation intensity of 142 is converted to a gamma value of 1.42, and the image processor immediately applies this parameter to adjust the red channel.

[0065] In practical applications, on an aluminum alloy inspection line, the blue channel image (resolution 6000×4000, 24 million pixels) was processed. A high-brightness threshold of 205 was set, and 2.88 million pixels with a brightness greater than 205 were found, accounting for 12%. A low-brightness threshold of 45 was set, and 4.32 million pixels with a brightness less than 45 were found, accounting for 18%. The basic light transmission reduction was calculated as 12% - 18% = -6%. This negative value indicates that there is light transmission loss in the blue channel.

[0066] Step 1052 queries the metal material database to determine that the aluminum alloy reflection influence factor is 1.15. The light transmission attenuation coefficient is calculated: taking the absolute value of the base attenuation of 6% and multiplying it by 1.15 yields 6.9%. An inverse relationship model is constructed: setting a baseline value of 180 and a magnification factor of 7, the formula "Compensation Intensity = 180 / (1 + 7 × Attenuation Coefficient)" is established. Substituting 6.9% into the calculation: 180 / (1 + 7 × 0.069) = 180 / 1.483 ≈ 121.4. This value represents the color compensation intensity required for the current blue channel.

[0067] In step 1053, the compensation intensity of 121.4 is converted into a pixel brightness conversion feature parameter: 121.4 / 100 = 1.214. This new parameter is transmitted to the blue channel adjustment module of the image processor via the communication interface. The processor immediately applies the gamma value of 1.214 to the current frame image, increasing the brightness of previously overly dark areas by 21.4% and correspondingly reducing the brightness of overly bright areas, thus significantly correcting the color difference caused by metallic reflections.

[0068] The overall solution in step 105 above establishes an intelligent reverse adjustment mechanism for light transmission loss and compensation intensity, thereby achieving dynamic optimization of pixel-level brightness parameters, effectively solving the color distortion problem under metallic reflective environments, and significantly improving the realism and stability of image color reproduction.

[0069] The following is a complete embodiment for steps 101-105: In step 101, on the automotive aluminum alloy wheel hub quality inspection line, an unmanned intelligent turntable performs rotational scanning at 10-degree intervals. When the turntable rotates to the 45-degree position, data acquisition is triggered simultaneously: the spectrum analyzer scans the wavelength distribution data of ambient light, recording 58 units of intensity for red light (600-780nm), 33 units for green light (500-600nm), and 9 units for blue light (380-500nm); the polarization sensor performs 10 samples within 200 milliseconds, recording the polarization angle change data as [30°, 45°, 30°, 30°, 60°, 30°, 45°, 30°, 30°, 30°]. When the turntable rotates to the 90-degree position, it acquires 22 units of red light, 48 units of green light, and 30 units of blue light, along with polarization data [60°, 75°, 60°, 75°, 60°, 75°, 60°, 75°, 60°, 75°, 60°, 75°], forming a raw angle-spectrum-polarization dataset.

[0070] Step 102 processes the data at the 45-degree position: the spectral data is divided into red, green, and blue light units; the polarization data within the red light unit [30°, 30°, 30°, 30°, 30°, 30°] appears 6 times in 10 samples → specifying the polarization direction as 30°; the polarization data in the green light unit [45°, 45°] appears 2 times → specifying the polarization direction as 45°; the polarization data in the blue light unit [60°] appears once → specifying the polarization direction as 60°; generating matrix entries: 45° position {red light unit → 30°, green light unit → 45°, blue light unit → 60°}. The process is repeated when the turntable is rotated to the 120-degree position to construct a complete dynamic matrix.

[0071] In step 103, when the turntable is detecting the stainless steel valve at a 60-degree position: the ambient light data shows that the blue light segment 380-500nm has the highest energy, with an intensity of 68 units × 120nm = 8160; the blue light unit at the 60-degree position in the query matrix corresponds to a target polarization angle of 80°; a binary command "01010000" / 80° encoding is sent to the programmable polarization modulation layer; the modulation layer applies a 4.2V voltage to drive the liquid crystal unit to rotate from 60° to 80°; the vibration direction of the transmitted light is synchronously adjusted to 80°, and the brightness of the highly reflective metal area is reduced by 40%.

[0072] Process the polarized image of the stainless steel part in step 104. For a 20-megapixel image: separate the green channel monochrome image, set the highlight threshold to 210, and count the pixels with brightness > 210: 1.8 million → 9%; set the shadow threshold to 35 for the same image, and count the pixels with brightness < 35: 3 million → 15%; calculate the basic light transmission reduction: 9% - 15% = -6%; stainless steel reflection correction factor 0.8 → light transmission attenuation coefficient = 6% × 0.8 = 4.8%.

[0073] In step 105, the light attenuation coefficient of the blue channel is 6.44%; input the inverse model: compensation intensity = 180 / (1+7×0.0644)≈124; generate pixel brightness conversion feature parameters: gamma value = 124 / 100 = 1.24; image processor updates parameters: the original gamma value of 1.0 is replaced with 1.24; apply the new parameters to process the current frame; the brightness of overly dark areas is increased by 24%, and the brightness of overly bright areas is reduced; the detection rate of scratches on the wheel hub surface is increased to 98%.

[0074] This solution achieves high-fidelity color reproduction in metallic reflective scenes by employing multi-angle spectral-polarization data collaborative acquisition on a turntable, dynamic matrix-driven polarization control, quantification analysis of channel transmittance loss, and reverse compensation model updates. This method overcomes the response hysteresis limitations of fixed polarization strategies, eliminates the processing delay defects of multi-frame fusion, and overcomes the material adaptation bottleneck of offline calibration. It simultaneously optimizes optical suppression and digital compensation during high-speed rotational detection, significantly improving defect identification accuracy and system robustness in industrial quality inspection scenarios.

[0075] Figure 3 This application provides a schematic diagram of the structure of an image color correction system for intelligent image processing, as shown in the embodiment of the present application. Figure 3 As shown, the system includes: The acquisition module 31 is used to simultaneously acquire ambient light wavelength distribution data and polarization angle change data corresponding to the rotation angle of the turntable when the unmanned intelligent turntable performs rotation scanning in an environment containing metallic reflective objects. The mapping module 32 is used to spatially map the ambient light wavelength distribution data and the polarization angle change data to generate a dynamic correspondence matrix between wavelength and polarization direction that varies with the turntable angle. The integrated module 33 is used to integrate a programmable polarization modulation layer in the optical path in front of the image sensor, and to call the dynamic correspondence matrix to drive the programmable polarization modulation layer to switch the transmission polarization direction in order to suppress the interference of the high reflectivity area of ​​the metal surface on the original image color data. The calculation module 34 is used to separate the red, green and blue channels of the image after polarization direction switching and calculate the channel light transmission attenuation coefficient caused by metal reflection based on the pixel intensity distribution of each channel after separation. The generation module 35 is used to establish an inverse correlation model between the light transmission attenuation coefficient and the color compensation intensity during the channel separation process. By inputting the light transmission attenuation coefficient into the inverse correlation model, the pixel brightness and darkness conversion feature parameters of each color channel are dynamically generated. By updating the pixel brightness and darkness conversion feature parameters, color restoration under metallic reflective environment is achieved.

[0076] Figure 3 The image color correction system for intelligent image processing described above can perform... Figure 1 The implementation principle and technical effects of the image color correction method for intelligent image processing described in the illustrated embodiment will not be repeated here. The specific methods by which each module and unit of the image color correction system for intelligent image processing in the above embodiments perform their operations have been described in detail in the embodiments related to this method, and will not be elaborated upon here.

[0077] In one possible design, Figure 3 The image color correction system for intelligent image processing shown in the embodiment can be implemented as a computing device, such as... Figure 4 As shown, the computing device may include a storage component 41 and a processing component 42; The storage component 41 stores one or more computer instructions, wherein the one or more computer instructions are invoked and executed by the processing component 42.

[0078] The processing component 42 is used for the above Figure 1 The embodiment describes an image color correction method for intelligent image processing.

[0079] The processing component 42 may include one or more processors to execute computer instructions to complete all or part of the steps in the above-described method. Alternatively, the processing component may be implemented as one or more application-specific integrated circuits (ASICs), digital signal processors (DSPs), digital signal processing devices (DSPDs), programmable logic devices (PLDs), field-programmable gate arrays (FPGAs), controllers, microcontrollers, microprocessors, or other electronic components to perform the above-described method.

[0080] Storage component 41 is configured to store various types of data to support operations at the terminal. The storage component can be implemented by any type of volatile or non-volatile storage device or a combination thereof, such as static random access memory (SRAM), electrically erasable programmable read-only memory (EEPROM), erasable programmable read-only memory (EPROM), programmable read-only memory (PROM), read-only memory (ROM), magnetic storage, flash memory, magnetic disk, or optical disk.

[0081] Of course, computing devices may also include other components, such as input / output interfaces, display components, communication components, etc.

[0082] Input / output interfaces provide interfaces between processing components and peripheral interface modules, which can be output devices, input devices, etc.

[0083] The communication components are configured to facilitate wired or wireless communication between computing devices and other devices.

[0084] The computing device can be a physical device or an elastic computing host provided by a cloud computing platform. In this case, the computing device can refer to a cloud server, and the aforementioned processing components, storage components, etc., can be basic server resources rented or purchased from the cloud computing platform.

[0085] This application also provides a computer storage medium storing a computer program, which, when executed by a computer, can perform the above-described functions. Figure 1 The illustrated embodiment is an image color correction method for intelligent image processing.

[0086] Those skilled in the art will clearly understand that, for the sake of convenience and brevity, the specific working processes of the systems, devices, and units described above can be referred to the corresponding processes in the foregoing method embodiments, and will not be repeated here.

[0087] The device embodiments described above are merely illustrative. The units described as separate components may or may not be physically separate. The components shown as units may or may not be physical units; that is, they may be located in one place or distributed across multiple network units. Some or all of the modules can be selected to achieve the purpose of this embodiment according to actual needs. Those skilled in the art can understand and implement this without any creative effort.

[0088] Through the above description of the embodiments, those skilled in the art can clearly understand that each embodiment can be implemented by means of software plus necessary general-purpose hardware platforms, and of course, it can also be implemented by hardware. Based on this understanding, the above technical solutions, in essence or the part that contributes to the prior art, can be embodied in the form of a software product. This computer software product can be stored in a computer-readable storage medium, such as ROM / RAM, magnetic disk, optical disk, etc., and includes several instructions to cause a computer device (which may be a personal computer, server, etc.) to execute the methods described in the various embodiments or some parts of the embodiments.

[0089] Finally, it should be noted that the above embodiments are only used to illustrate the technical solutions of this application, and are not intended to limit them. Although this application has been described in detail with reference to the foregoing embodiments, those skilled in the art should understand that modifications can still be made to the technical solutions described in the foregoing embodiments, or equivalent substitutions can be made to some of the technical features. Such modifications or substitutions do not cause the essence of the corresponding technical solutions to deviate from the spirit and scope of the technical solutions of the embodiments of this application.

Claims

1. An image color correction method for intelligent image processing, characterized in that, include: When the unmanned intelligent turntable performs rotational scanning in an environment containing metallic reflective objects, it simultaneously collects ambient light wavelength distribution data and polarization angle change data corresponding to the turntable rotation angle. The ambient light wavelength distribution data and polarization angle change data are spatially mapped to generate a dynamic correspondence matrix between wavelength and polarization direction that varies with the turntable angle. A programmable polarization modulation layer is integrated in the optical path in front of the image sensor. The dynamic correspondence matrix is ​​called to drive the programmable polarization modulation layer to switch the transmission polarization direction, so as to suppress the interference of the high reflectivity area of ​​the metal surface on the original image color data. The red, green and blue channels of the image after polarization direction switching are separated, and the light transmission attenuation coefficient caused by metal reflection is calculated based on the pixel intensity distribution of each channel after separation. A reverse correlation model is established between the light transmission attenuation coefficient and the color compensation intensity during the channel separation process. By inputting the light transmission attenuation coefficient into the reverse correlation model, pixel brightness and darkness conversion feature parameters of each color channel are dynamically generated. Color restoration under metallic reflective environment is achieved by updating the pixel brightness and darkness conversion feature parameters.

2. The method according to claim 1, characterized in that, The step of spatially mapping the ambient light wavelength distribution data with the polarization angle change data to generate a dynamic correspondence matrix between wavelength and polarization direction that varies with the turntable angle includes: A spatial position sequence of the turntable rotation angle is set, and ambient light wavelength distribution data and polarization angle change data are synchronously collected at each spatial position in the spatial position sequence. The ambient light wavelength distribution data at the same spatial location are spatially allocated according to the spectral range to form spectral spatial units, and the highest frequency value of the polarization angle change data in the spectral spatial unit is taken as the specified polarization direction. A spatial correspondence is established between the spectral spatial units and the specified polarization direction, generating a dynamic correspondence matrix between wavelength and polarization direction that varies with the turntable angle.

3. The method according to claim 1, characterized in that, The method of integrating a programmable polarization modulation layer in the optical path in front of the image sensor, and using the dynamic correspondence matrix to drive the programmable polarization modulation layer to switch the transmission polarization direction to suppress the interference of highly reflective areas on the metal surface on the original image color data, includes: The dominant spectral interval is extracted from the ambient light wavelength distribution data, and the target polarization angle is obtained by querying the dynamic correspondence matrix based on the dominant spectral interval. A programmable polarization modulation layer is integrated in the optical path in front of the image sensor, and a drive command carrying the target polarization angle is sent to the programmable polarization modulation layer. According to the driving command, the physical arrangement direction of the internal light adjustment plate is adjusted through the programmable polarization modulation layer so that the transmission polarization direction is aligned with the target polarization angle, thereby suppressing the interference of the high reflectivity area of ​​the metal surface on the original image color data.

4. The method according to claim 1, characterized in that, The process involves establishing an inverse correlation model between the light transmittance attenuation coefficient and the color compensation intensity during channel separation. This model dynamically generates pixel brightness transition feature parameters for each color channel by inputting the light transmittance attenuation coefficient into the inverse correlation model. Updating these pixel brightness transition feature parameters enables color restoration under metallic reflective environments. Extract the proportion of high-value regions and low-value regions of pixel intensity during the channel separation process, and calculate the difference between the proportion of high-value regions and the proportion of low-value regions as the basic light transmission reduction amount; Multiply the basic light transmission reduction by the metal reflection influence factor to output the light transmission attenuation coefficient, and construct an inverse correlation model between the light transmission attenuation coefficient and the color compensation intensity during the channel separation process. The light transmission attenuation coefficient is input into the inverse correlation model, and the color compensation intensity value is output. The pixel brightness and darkness conversion feature parameters of each color channel are generated based on the color compensation intensity value. Color restoration under metallic reflective environment is achieved by updating the pixel brightness and darkness conversion feature parameters.

5. The method according to claim 1, characterized in that, The process of separating the red, green, and blue channels of the image after polarization direction switching, and calculating the channel transmittance attenuation coefficient caused by metal reflection based on the pixel intensity distribution of each channel after separation, includes: Color component decomposition is performed on the image after polarization direction switching, and the red component image, green component image and blue component image are extracted respectively. The proportion of the number of pixels with brightness values ​​exceeding the set brightness threshold in each color component image is counted as the proportion of bright pixels. The proportion of dark pixels is calculated as the percentage of pixels in an image with the same color component whose brightness value is lower than a set dark threshold. The numerical difference between the proportion of bright pixels and the proportion of dark pixels is calculated as the base value of channel light transmission loss. The base value of channel light transmission loss is multiplied by the metal reflection correction factor to output the channel light transmission attenuation coefficient caused by metal reflection.

6. The method according to claim 3, characterized in that, The step of extracting the dominant spectral interval from the ambient light wavelength distribution data and obtaining the target polarization angle by querying the dynamic correspondence matrix based on the dominant spectral interval includes: The ambient light wavelength distribution data is divided into multiple spectral segments according to fixed wavelength intervals, and the sum of the energy values ​​corresponding to all wavelength points in each spectral segment is calculated as the energy concentration degree. The spectral segment containing the largest energy concentration is selected as the dominant spectral segment; Search the dynamic correspondence matrix for record entries that perfectly match the wavelength range of the dominant spectral region, and extract the corresponding polarization direction record from the record entries as the target polarization angle.

7. The method according to claim 2, characterized in that, The step of spatially allocating the ambient light wavelength distribution data according to the spectral range to form spectral spatial units at the same spatial location, and using the highest frequency value of the polarization angle change data in the spectral spatial unit as the specified polarization direction, includes: In the same spatial location, the ambient light wavelength distribution data is divided into spectral spatial units according to the progressive relationship of wavelength values ​​corresponding to the spectral range from small to large. Within the wavelength range covered by the spectral spatial unit, statistically analyze the corresponding polarization angle variation data and record the total number of occurrences of the polarization angle variation data within the wavelength range; The total number of occurrences is compared, and the polarization angle change data with the highest total number of occurrences is selected as the specified polarization direction of the spectral spatial unit.

8. An image color correction system for intelligent image processing, characterized in that, include: The acquisition module is used to simultaneously acquire ambient light wavelength distribution data and polarization angle change data corresponding to the rotation angle of the turntable when the unmanned intelligent turntable performs rotational scanning in an environment containing metallic reflective objects. The mapping module is used to spatially map the ambient light wavelength distribution data with the polarization angle change data to generate a dynamic correspondence matrix between wavelength and polarization direction that varies with the turntable angle. An integrated module is used to integrate a programmable polarization modulation layer in the optical path in front of the image sensor, and to call the dynamic correspondence matrix to drive the programmable polarization modulation layer to switch the transmission polarization direction in order to suppress the interference of the high reflectivity area of ​​the metal surface on the original image color data. The calculation module is used to separate the red, green and blue channels of the image after polarization direction switching, and calculate the channel light transmission attenuation coefficient caused by metal reflection based on the pixel intensity distribution of each channel after separation. The generation module is used to establish an inverse correlation model between the light transmission attenuation coefficient and the color compensation intensity during the channel separation process. By inputting the light transmission attenuation coefficient into the inverse correlation model, the pixel brightness and darkness conversion feature parameters of each color channel are dynamically generated. By updating the pixel brightness and darkness conversion feature parameters, color restoration under metallic reflective environment is achieved.

9. A computing device, characterized in that, It includes a processing component and a storage component; the storage component stores one or more computer instructions; the one or more computer instructions are invoked and executed by the processing component to implement an image color correction method for intelligent image processing as described in any one of claims 1 to 7.

10. A computer storage medium, characterized in that, The device contains a computer program that, when executed by a computer, implements an image color correction method for intelligent image processing as described in any one of claims 1 to 7.

Images