Method and system for color separation of digital printing images based on tone mapping

Abstract

The application discloses a color separation processing method and system for digital printing images based on tone mapping, and relates to the technical field of image processing. The method comprises the following steps: loading original RGB color data of a source high dynamic range image, converting the original RGB color data to a CIELAB space to obtain original luminance components and original chroma components of each pixel; calculating a local contrast parameter and a saturation parameter to generate a weight map for representing coupling strength of tone mapping and gamut mapping; constructing a joint objective function comprising a tone fidelity term, a gamut adaptation term and a coupling constraint term; solving the joint objective function by using an iterative optimization algorithm, alternately optimizing a tone mapping parameter vector and a gamut mapping parameter vector in each iteration, and performing feedback correction based on the weight map; and applying the optimal parameter vector to the source image to obtain multi-channel ink values after color separation and outputting the multi-channel ink values. The application solves the technical problem of cumulative distortion caused by serial processing of tone mapping and gamut mapping in the prior art.

Description

Technical Field

[0001] This invention relates to the field of image processing technology, and more specifically to a method and system for color separation processing of digital printing images based on tone mapping. Background Technology

[0002] In the field of digital printing image color separation processing, high dynamic range source images must be compressed to the standard dynamic range of the printing equipment through tone mapping, and then adapted to the color gamut boundary of the printing equipment through color gamut mapping.

[0003] In existing technologies, the two steps mentioned above are typically executed sequentially. For example, Dolby Laboratories' patent CN106204474B proposes a local multi-level tone mapping operator that optimizes tone mapping effects through multi-scale weighted recombination, but it only involves the internal optimization of a single mapping process. Similarly, Konica Minolta's patent CN104717432B proposes a method for selecting tone mapping operators and gamut mapping algorithms based on scene classification, but it still maintains an architecture where the two are processed independently, making improvements only at the level of selecting different algorithms.

[0004] However, the aforementioned sequential processing mode has inherent drawbacks: after tone mapping alters the brightness distribution of the original image, colors originally located within the color gamut may extend beyond the gamut boundaries due to brightness changes, forcing subsequent gamut mapping to undergo additional compression and resulting in cumulative distortion. Currently, there is a lack of a color separation method that can jointly optimize tone mapping and gamut mapping. Summary of the Invention

[0005] This invention provides a method and system for color separation processing of digital printing images based on tone mapping, which solves the technical problem of cumulative distortion caused by the serial processing of tone mapping and color gamut mapping in the prior art.

[0006] In a first aspect, the present invention provides a color separation processing method for digital printing images based on tone mapping, the method comprising:

[0007] Load the original RGB color data of the source high dynamic range image, and convert the original RGB color data to CIELAB space to obtain the original luminance component and original chrominance component of each pixel;

[0008] Local contrast parameters are calculated based on the original luminance components, saturation parameters are calculated based on the original chrominance components, and a weight map is generated based on the local contrast parameters and the saturation parameters. The weight map is used to characterize the coupling strength between tone mapping and gamut mapping at each pixel position.

[0009] Construct a joint objective function, which includes at least a hue fidelity term, a gamut adaptation term, and a coupling constraint term;

[0010] The joint objective function is solved by an iterative optimization algorithm. Each iteration includes: alternately optimizing the tone mapping parameter vector and the gamut mapping parameter vector, and making feedback corrections to the two types of parameters based on the weight graph.

[0011] The optimal tone mapping parameter vector and the optimal gamut mapping parameter vector obtained when the iteration termination condition is met are applied to the source high dynamic range image to obtain the multi-channel ink volume values ​​after color separation and output them.

[0012] Secondly, the present invention also provides a color separation processing system for digital printing images based on tone mapping, the system comprising:

[0013] The color space conversion module is used to load the original RGB color data of the source high dynamic range image and convert the original RGB color data to CIELAB space to obtain the original luminance component and original chrominance component of each pixel.

[0014] The weight map generation module is used to calculate local contrast parameters based on the original luminance components, calculate saturation parameters based on the original chrominance components, and generate a weight map based on the local contrast parameters and the saturation parameters. The weight map is used to characterize the coupling strength between tone mapping and gamut mapping at each pixel position.

[0015] A joint objective function construction module is used to construct a joint objective function, which includes at least a hue fidelity term, a gamut adaptation term, and a coupling constraint term.

[0016] The iterative optimization solution module is used to solve the joint objective function through an iterative optimization algorithm. Each iteration includes: alternately optimizing the tone mapping parameter vector and the gamut mapping parameter vector, and performing feedback correction on the two types of parameters based on the weight graph.

[0017] The color separation output module is used to apply the optimal tone mapping parameter vector and the optimal color gamut mapping parameter vector obtained when the iteration termination condition is met to the source high dynamic range image, obtain the multi-channel ink volume value after color separation and output it.

[0018] One or more technical solutions provided in this invention have at least the following technical effects or advantages:

[0019] First, this invention generates a weighted graph based on local contrast and saturation parameters, dynamically adjusting the coupling constraint strength between the tone mapping parameter vector and the gamut mapping parameter vector during iterative optimization. In edge and textured regions, where local contrast is high and saturation is low, the weight values ​​are large, strengthening the coupling constraint. In these regions, tone mapping and gamut mapping are highly coordinated, avoiding detail distortion caused by brightness compression followed by gamut clipping in sequential processing. In flat regions, where local contrast is low and saturation is high, the weight values ​​are small, relaxing the coupling constraint and allowing gamut mapping greater independent adjustment space to ensure color accuracy. Compared to existing technologies where tone mapping and gamut mapping are executed sequentially and independently without feedback mechanisms, the weighted graph-driven coupling constraint enables coordinated adjustment during optimization, reducing cumulative distortion.

[0020] Second, this invention incorporates tone mapping and gamut mapping into a unified optimization framework by constructing a joint objective function that includes tone fidelity, gamut adaptation, and coupling constraint terms. Through iterative and alternating optimization of the two types of parameter vectors, each update of the tone mapping parameters affects the search direction and step size of the gamut mapping parameters, and each update of the gamut mapping parameters, in turn, influences the starting point for the optimization of the tone mapping parameters in the next round. Compared to existing technologies that only improve the selection of different algorithms while maintaining an independent processing architecture, this invention breaks down the sequential barrier between the two at the objective function level, enabling tone mapping and gamut mapping to guide each other and converge collaboratively based on the current optimal intermediate results in each iteration. Attached Figure Description

[0021] To more clearly illustrate the technical solutions in the embodiments of the present invention, the accompanying drawings used in the description of the embodiments will be briefly introduced below. Obviously, the accompanying drawings described below are only some embodiments of the present invention. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0022] Figure 1 This is a flowchart illustrating the color separation processing method for digital printing images based on tone mapping provided in an embodiment of the present invention.

[0023] Figure 2 This is a logic diagram of a digital printing image color separation processing method based on tone mapping provided in an embodiment of the present invention;

[0024] Figure 3 This is an example data table diagram of weight map generation for the color separation processing method for digital printing images based on tone mapping provided in this embodiment of the invention;

[0025] Figure 4 This is a schematic diagram of the structure of a digital printing image color separation processing system based on tone mapping provided in an embodiment of the present invention;

[0026] The diagram is labeled as follows: Color space conversion module 11, weight graph generation module 12, joint objective function construction module 13, iterative optimization solution module 14, and color separation output module 15. Detailed Implementation

[0027] The technical solutions of the present invention will be clearly and completely described below with reference to the accompanying drawings in the embodiments of the present invention.

[0028] Example 1, as Figure 1 As shown, this invention provides a flowchart illustrating a color separation processing method for digital printing images based on tone mapping; as... Figure 2 As shown, this invention provides a scheme logic diagram for a color separation processing method for digital printing images based on tone mapping. The method includes:

[0029] S100: Load the original RGB color data of the source high dynamic range image, convert the original RGB color data to CIELAB space, and obtain the original luminance component and original chrominance component of each pixel;

[0030] High dynamic range (HDR) images are stored in RGB format, with a brightness dynamic range typically reaching tens of thousands to one, far exceeding the hundreds to one dynamic range of digital printing equipment. The three channels in the RGB space simultaneously encode luminance and chrominance information. Directly compressing luminance will consequently alter chrominance, leading to inaccurate subsequent color gamut mapping. The CIELAB space decomposes color into a luminance component L and two chrominance components a and b. Luminance and chrominance are numerically independent, and the perceptual difference between the two colors can be approximated by geometric distance, providing a device-independent unified reference for tone mapping and color gamut mapping. The luminance component contains both low-frequency information such as overall illumination distribution and light-dark structure, as well as high-frequency details such as edges and textures. Applying equal global compression to both components will weaken local details along with the dynamic range.

[0031] Step S100 provided in this embodiment of the invention further includes:

[0032] Perform layer decomposition on the original luminance components to obtain a base layer representing the overall luminance distribution and a detail layer representing local texture details;

[0033] The joint objective function also includes a detail retention term, which is constructed based on the differential compression constraints between the base layer and the detail layer, and is used to characterize the balance between overall brightness compression and local detail retention during tone mapping.

[0034] The specific implementation method is as follows:

[0035] First, the raw RGB color data of the source high dynamic range image is loaded. The source high dynamic range image is stored as a digital image file, and the file format can be an image format that supports high dynamic range, such as OpenEXR or TIFF. The raw RGB color data is represented by the R, G, and B channel values ​​for each pixel. The bit depth of each channel is determined by the acquisition device or generation method of the source image, and is usually a 16-bit or 32-bit floating-point number, capable of recording a wide brightness range from extremely dark to extremely bright. The loaded raw RGB color data is stored in memory as a three-dimensional array. The three dimensions correspond to the width, height, and number of color channels of the image, respectively. Each element is a floating-point number representing the intensity value of that pixel in that color channel. The value range is not limited to 0 to 1 and can include values ​​far exceeding the standard dynamic range.

[0036] Then, the source high dynamic range image is converted from RGB space to CIELAB space. The conversion process consists of two stages: the first stage converts the RGB values ​​into CIE XYZ tristimulus values ​​after linearization and inverse Gamma correction; the second stage converts the XYZ tristimulus values ​​into three components, L, a, and b, using the non-linear conversion formula specified by CIE. After the conversion, the original luminance component L and the original chrominance components a and b are obtained for each pixel. The value of the original luminance component L is typically between 0 and 100.

[0037] Furthermore, a hierarchical decomposition is performed on the original luminance components. This hierarchical decomposition employs an edge-preserving filter to smooth the original luminance components. The edge-preserving filter assigns weighted contributions to each neighboring pixel by calculating the similarity in both the spatial and numerical domains between each neighboring pixel and the center pixel within the filtering window. Neighboring pixels that are spatially closer to the center pixel and have smaller luminance value differences receive higher weights; neighboring pixels with larger luminance value differences from the center pixel, even if spatially close, have their weights suppressed. Through this weighting mechanism, the filter exhibits a strong smoothing effect in flat areas of the image; in areas with drastic luminance changes such as edges, the filter smooths independently on both sides of the edge, avoiding cross-boundary blurring.

[0038] The output of the edge-preserving filter forms the base layer, representing the low-frequency information retained after smoothing, reflecting large-scale brightness variations in the scene determined by illumination distribution and object geometry. Subtracting the base layer from the original luminance component creates the detail layer, representing the high-frequency information filtered out by smoothing, reflecting surface texture, fine edges, and local contrast variations. The parameters for layer decomposition are set based on the noise level of the source image and the target printing resolution. The radius of the filter window determines the range of spatial domain smoothing; a larger radius results in a smoother base layer. The similarity threshold in the value domain determines the sensitivity of smoothing in the luminance domain; a larger threshold ensures the base layer tracks luminance changes more closely.

[0039] After obtaining the base layer and detail layer, the joint objective function in subsequent steps will include a detail preservation term. This detail preservation term applies compression constraints of varying strengths to the base and detail layers, ensuring that the joint objective function compresses the overall dynamic range while preserving local texture and edge information. The base layer carries the overall brightness distribution, and its compression strength is determined based on the dynamic range of the printing equipment; the detail layer carries local texture, and its compression strength is set independently to ensure that local details can still be reproduced by the printing equipment after overall dynamic range compression.

[0040] The following technical effects were achieved through this step:

[0041] First, this step converts the source high dynamic range image from RGB space to CIELAB space, numerically separating the luminance and chrominance components. Compared to direct processing in RGB space, the luminance and chrominance components are independent in CIELAB space. Subsequent steps can perform tone mapping on the luminance component and gamut mapping on the chrominance component separately, without interference between the two operations. At the same time, the perceptual uniformity of CIELAB space ensures consistency between the calculated distance between the two colors and the difference perceived by the human eye.

[0042] Second, this step performs hierarchical decomposition on the original luminance components using an edge-preserving filter to obtain a base layer and a detail layer. Compared to a scheme that directly performs global compression on the luminance components, this method separates the low-frequency information of the overall luminance distribution and the high-frequency information of the local texture into two independent layers. This provides hierarchical input for applying differential compression constraints to the base layer and detail layer in the subsequent joint objective function, enabling tone mapping to preserve local details while compressing the dynamic range.

[0043] S200: Calculate local contrast parameters based on the original luminance components, calculate saturation parameters based on the original chrominance components, and generate a weight map based on the local contrast parameters and the saturation parameters. The weight map is used to characterize the coupling strength between tone mapping and gamut mapping at each pixel position.

[0044] In high dynamic range (HDR) images, the coupling requirements for tone mapping and gamut mapping vary significantly across different spatial locations. At object edges and in textured areas, where local contrast is high, compression of the luminance component directly impacts the visibility of details in these areas. Gamut mapping should be tightly coupled with tone mapping to avoid color shifts or detail loss in the chrominance components of these areas due to independent mapping after luminance compression. In large, flat, uniform color areas with low local contrast and high saturation, gamut mapping requires a larger independent adjustment space to accurately map colors outside the printing device's color gamut into the gamut. It should not be overly constrained by tone mapping. If all pixel locations use the same coupling strength, insufficient coupling in detail areas may lead to color distortion, while excessive coupling in flat areas may result in decreased color accuracy.

[0045] Step S200 provided in this embodiment of the invention includes:

[0046] Generate a weighted graph, including:

[0047] Based on the local contrast parameter and the saturation parameter, the normalized values ​​of the local contrast parameter and the normalized values ​​of the saturation parameter are weighted and summed according to a preset weighting coefficient to calculate the weight value of each pixel.

[0048] Wherein, the normalized value of the local contrast parameter is the ratio of the local contrast parameter to the sum of the maximum and minimum local contrast parameters in the entire image, and the normalized value of the saturation parameter is the ratio of the saturation parameter to the sum of the maximum and minimum saturation parameters in the entire image.

[0049] The specific implementation method is as follows:

[0050] First, based on the original luminance components, the local contrast parameter of each pixel is calculated. The local contrast parameter measures the drastic change in brightness of a pixel within its surrounding neighborhood. For each pixel in the image, a local window of a preset size is taken centered on that pixel. The window size is determined based on the image resolution and texture scale; a window that is too small will cause noise interference in the contrast calculation, while a window that is too large will result in over-smoothing of local characteristics. The preset window size is determined by combining the resolution of the digital printed image and the local perceptual characteristics of the human visual system. At a typical printing resolution of 300 dpi, the window size is 15 pixels by 15 pixels. This size corresponds to an image area of ​​approximately 1.3 mm square, matching the smallest detail size that the human eye can distinguish at a normal reading distance. The window size can be determined through pre-experimentation. For example, several HDR images with different texture complexities are selected, and the correlation coefficient between the local contrast parameter and subjective texture saliency is calculated for each window size. The window size corresponding to the maximum correlation coefficient is selected as the preset size. Within the local window, the standard deviation of the luminance values ​​of all pixels is calculated, and this standard deviation is used as the local contrast parameter for that pixel. The physical meaning of standard deviation is the average deviation of the brightness value within a window from the mean. The larger the standard deviation, the more drastic the brightness change within the window, and the higher the probability that the pixel is located in an edge or textured area; the smaller the standard deviation, the more uniform the brightness within the window, and the higher the probability that the pixel is located in a flat area.

[0051] Based on the original chroma components a and b, the saturation parameter of each pixel is calculated. The saturation parameter measures the vividness of a pixel's color. It is calculated as follows: the saturation parameter equals the square root of the sum of the squares of the a and b components of that pixel. The a and b components represent the degree of color deviation in the green-red and blue-yellow directions, respectively. The square root of the sum of their squares is the Euclidean distance of that color from the neutral gray axis; a larger distance indicates a more vivid color and higher saturation.

[0052] Then, the local contrast and saturation parameters are normalized separately. The purpose of normalization is to map two parameters with different dimensions and value ranges to the same numerical scale, making them comparable when weighted summed.

[0053] The normalization method for local contrast parameters is as follows: traverse all pixels in the entire image and find the maximum and minimum values ​​of the local contrast parameters. For each pixel, the normalized value of its local contrast parameter is equal to the local contrast parameter of that pixel divided by the sum of the maximum and minimum values ​​of local contrast parameters in the entire image. The denominator uses the sum of the maximum and minimum values ​​instead of their difference. This is to ensure that the normalized value is always positive and the numerator does not exceed the denominator, mapping the normalized result to between zero and one. Compared to dividing by the maximum value, including the minimum value in the denominator reduces the sensitivity of the normalized value to isolated noise points. The normalized value of the saturation parameter uses the same denominator as the normalized value of the local contrast parameter: the saturation parameter of that pixel divided by the sum of the maximum and minimum values ​​of the saturation parameters in the entire image.

[0054] Finally, the normalized values ​​of the local contrast and saturation parameters are summed according to preset weighting coefficients to calculate the weight value of each pixel. The weight value characterizes the coupling strength requirement between tone mapping and gamut mapping at that pixel location. The weighting coefficients are set based on the fact that local contrast reflects areas sensitive to brightness changes; pixels with higher normalized values ​​require more protection of local details in tone mapping and have a stronger coupling requirement between gamut mapping and tone mapping, thus receiving a larger weighting coefficient. Saturation reflects the core requirement of gamut mapping; pixels with higher normalized values ​​require more independent and accurate gamut mapping and have a weaker dependence on coupling constraints, thus receiving a smaller weighting coefficient. A larger weight value after weighted summation indicates a higher coupling strength requirement between tone mapping and gamut mapping at that pixel location, resulting in stronger constraints on the coupling term at that pixel location in subsequent iterations. Conversely, a smaller weight value indicates a lower coupling strength requirement at that pixel location, allowing for greater independent adjustment space for gamut mapping at that pixel location.

[0055] After performing the weighted summation calculation on all pixels of the entire image, each pixel receives a weight value. The weight values ​​of all pixels are arranged according to their spatial positions to form a two-dimensional matrix with the same resolution as the original image; this two-dimensional matrix is ​​the weight map. The weight map is stored as an image, where each element corresponds to a pixel position in the image, and the element's value is the weight value for that pixel. The range of its value is determined by the range of normalized values ​​and the setting of the weighting coefficients. The weight map is used to characterize the coupling strength requirement between tone mapping and gamut mapping at each pixel position.

[0056] Step S200 provided in this embodiment of the invention further includes:

[0057] Feedback correction is performed on the tone mapping parameter vector and the gamut mapping parameter vector based on the weight map, including:

[0058] For pixel positions in the weighted graph where the weight value is greater than a preset high weight threshold, the coefficient of the coupling constraint term is increased in the next iteration;

[0059] For pixel positions in the weighted graph where the weight value is less than a preset low weight threshold, the coefficient of the coupling constraint term is reduced in the next iteration.

[0060] The weight graph is updated synchronously in each iteration, and the weight graph is replaced with an updated weight graph recalculated based on the intermediate results obtained from the current iteration.

[0061] The specific implementation method is as follows:

[0062] In each round of iterative optimization, the weight value of each pixel in the current weight map is compared with preset high-weight thresholds and low-weight thresholds. The high-weight thresholds and low-weight thresholds are determined based on the statistical distribution of the weight map. The high-weight threshold is used to filter out detailed regions with high local contrast and low saturation, while the low-weight threshold is used to filter out flat regions with low local contrast and high saturation. The high-weight thresholds and low-weight thresholds are preset by taking the upper and lower quantiles of the statistical distribution of the initial weight map, respectively. For example, the high-weight threshold is taken as the 70th percentile of all pixel weight values ​​in the initial weight map, indicating that pixels with weight values ​​higher than this threshold are high-demand regions in the top 30% of the entire image; the low-weight threshold is taken as the 30th percentile, indicating that pixels with weight values ​​lower than this threshold are low-demand regions in the bottom 30% of the entire image. When the weight values ​​are concentrated, the proportion of pixels in the middle region between the upper and lower thresholds is relatively high, and the coupling constraint coefficient remains unchanged for most pixels. When the weight values ​​are dispersed, more pixels are assigned to the high-weight or low-weight regions at both ends, expanding the range of differential adjustment and adapting to the dispersion of the image's coupling requirements.

[0063] For pixel locations where the weight value in the weight map exceeds a preset high-weight threshold, the coefficients of the coupling constraint term are increased in the next iteration. These pixel locations are situated in the image's edges and texture regions, rich in local detail, requiring high protection of detail visibility during tone mapping. Adjusting the tone mapping parameters directly alters the brightness and chromaticity relationship in these regions, and the gamut mapping parameters must closely follow these tone mapping parameter changes to ensure accurate color even after brightness compression. Increasing the coefficients of the coupling constraint term means that at this pixel location, the penalty weight of the coupling constraint term in the joint objective function is amplified. The tone mapping parameters and gamut mapping parameters are forced to maintain a more consistent adjustment pace during optimization, resulting in high synergy between the two in this region and mitigating color deviations caused by independent adjustments.

[0064] For pixel locations where the weight value in the weighted graph is less than a preset low-weight threshold, the coefficient of the coupling constraint term is reduced in the next iteration. These pixel locations are situated in large, flat areas with gradual local brightness changes, making them insensitive to brightness compression. Simultaneously, flat areas typically have higher saturation, requiring greater accuracy in gamut mapping. The adjustment range of the luminance component in tone mapping is smaller in such areas, allowing the gamut mapping parameters to independently search for the optimal gamut adaptation direction with less coupling constraint. Reducing the coefficient of the coupling constraint term means that at that pixel location, the penalty weight of the coupling constraint term in the joint objective function is reduced, granting the gamut mapping parameters greater independent adjustment freedom to more closely approximate the gamut boundary of the printing equipment and ensure high-fidelity color reproduction in flat areas.

[0065] For pixel locations with weight values ​​between the high and low weight thresholds, the coefficients of the coupling constraint term remain unchanged. These regions occupy an intermediate transitional state between the requirements for detail preservation and color gamut mapping accuracy, and maintaining the existing coupling strength achieves a good balance.

[0066] After completing the above feedback correction in each iteration, the coupling constraint coefficients are adjusted differently for different pixel positions based on the weight map. The optimization direction of tone mapping parameters and gamut mapping parameters is brought closer by coupling constraints in the detail area and relaxed by coupling constraints in the flat area.

[0067] Step S200 provided in this embodiment of the invention further includes:

[0068] The weight map is updated after each iteration based on the intermediate optimization results of the current iteration. Specifically, based on the intermediate luminance and intermediate chrominance components obtained in this iteration, the local contrast and saturation parameters of each pixel are recalculated, and a new weight map is generated for the next iteration.

[0069] The specific implementation method is as follows:

[0070] The initial iteration uses an initial weight map calculated based on the original image data. After the initial iteration, the joint objective function is solved to obtain the intermediate luminance and intermediate chrominance components after the first round of optimization. The intermediate luminance component undergoes initial tone mapping processing, and its dynamic range of luminance has been initially compressed, the contrast of some high-contrast areas has been reduced, and the local contrast parameter has decreased accordingly. The intermediate chrominance component undergoes initial color gamut mapping processing, and the chrominance values ​​of some high-saturation pixels have been initially corrected and compressed into the printing color gamut, and the saturation parameter has changed accordingly. If the initial weight map calculated based on the original image data continues to drive subsequent iterations, the weight map reflects the coupling requirement distribution before optimization, which deviates from the actual state of the image after initial optimization.

[0071] Therefore, after each iteration, based on the intermediate luminance and intermediate chrominance components obtained in this round, the local contrast and saturation parameters of each pixel are recalculated. The recalculation method is the same as the steps described above: the local contrast parameters are calculated by replacing the original luminance components with the intermediate luminance components, and the saturation parameters are calculated by replacing the original chrominance components with the intermediate chrominance components. Then, the two parameters are normalized separately, using the same normalization method as described above, by summing them according to preset weighting coefficients to generate a new weight map for the next iteration. The new weight map replaces the old weight map and serves as the basis for feedback correction in the next iteration.

[0072] The weight map transforms from a static distribution of the original image into parameters that dynamically iterate with the optimization process. After each optimization round, the brightness and chromaticity distributions of the intermediate results move closer to the optimal state, and the new weight map generated based on these intermediate results also moves closer to the optimal coupling distribution. The feedback correction is synchronized with the current state of the controlled object, ensuring that the adjustment of the coupling constraint coefficients always aligns with the current optimization level. As iterations proceed, the intermediate brightness and chromaticity components gradually converge to a better state, and the weight map gradually approaches the optimal weight distribution, driving continuous and coordinated adjustment of tone mapping and gamut mapping during the optimization process.

[0073] For example, Figure 3 This is an example data table diagram of weighted map generation provided in this embodiment of the invention. The diagram illustrates the calculation process of the weighted map at four representative pixel locations in different region types. The pixel location is the row and column coordinates of the pixel in the image. Region types are divided into four types: edge texture region, flat region, boundary transition region, and high-saturation texture region, each corresponding to different coupling requirements. The original brightness L, original chromaticity a, and b are obtained in step S100. The local contrast parameter is the standard deviation of the brightness value within the local window of the pixel, and the saturation parameter is the square root of the sum of the squares of the a and b components. The contrast normalization value is the local contrast parameter divided by the sum of the maximum and minimum local contrast values ​​of the entire image, and the saturation normalization value is the saturation parameter divided by the sum of the maximum and minimum saturation values ​​of the entire image. The weight values ​​are obtained by weighting and summing according to preset weighting coefficients, i.e., a contrast weight of 0.7 and a saturation weight of 0.3.

[0074] The following technical effects were achieved through this step:

[0075] First, this step generates a weight map by calculating and weighting local contrast and saturation parameters, thus quantifying the coupling strength requirement between tone mapping and gamut mapping at each pixel location into a comparable value. Detailed areas with high local contrast and low saturation receive a larger weight, while flat areas with low local contrast and high saturation receive a smaller weight. Through a differentiated coupling strategy driven by the weight map, a pixel-level adaptive coupling strength allocation mechanism is established between tone mapping and gamut mapping.

[0076] Second, this step differentiates and adjusts the coupling constraint coefficients based on the weighted map. High-weight pixels enhance the coupling constraint coefficients, ensuring high coordination between tone mapping and gamut mapping parameters in that region. Brightness compression and gamut adaptation occur simultaneously, mitigating the color distortion in detail areas caused by insufficient coupling in gamut mapping after changes in brightness distribution. Low-weight pixels reduce the coupling constraint coefficients, allowing gamut mapping parameters greater independent adjustment space in that region to more closely approximate the gamut boundaries of the printing equipment.

[0077] Third, this step dynamically updates the weight map based on the current intermediate optimization results in each iteration, allowing the weight distribution to adaptively adjust as the optimization progresses. Compared to schemes that rely solely on a fixed initial weight map to drive the iteration process, this approach continuously regenerates the weight map based on the actual states of the intermediate luminance and chrominance components during optimization. The coupling strength distribution adapts to the current iteration, creating a closed-loop feedback between the weight map and the optimization progress. This ensures that the coupling strength adjustment gradually approaches the optimal distribution as the optimization converges.

[0078] S300: Construct a joint objective function, which includes at least a hue fidelity term, a gamut adaptation term, and a coupling constraint term;

[0079] The above steps generate a weight map, quantifying the required coupling strength between tone mapping and gamut mapping at each pixel location. In the traditional serial processing mode, tone mapping first compresses the luminance component, and gamut mapping then adjusts the compressed chrominance component. When compressing luminance, tone mapping does not consider the gamut mapping requirements of the region, which may cause colors originally within the printing gamut to exceed the gamut boundary due to luminance changes. Similarly, when adjusting chrominance, gamut mapping does not consider the degree of luminance compression in the region, and may excessively sacrifice color saturation when cropping or compressing colors exceeding the gamut boundary. The two processes are optimized independently, lacking coordination, ultimately resulting in cumulative distortion in the output image.

[0080] The specific implementation method is as follows:

[0081] First, a tone fidelity term is constructed. This term measures the deviation between the luminance component after tone mapping and the original luminance component. Its purpose is to constrain tone mapping to maintain the structural characteristics of the original luminance distribution as much as possible while compressing the overall dynamic range. The tone fidelity term is calculated as follows: for all pixels in the image, the square of the difference between the mapped luminance value and the original luminance value is calculated and summed over the entire image. For the base layer and detail layer obtained from the layered decomposition in step S100, the tone fidelity term calculates the mapping deviation of the base layer and the mapping deviation of the detail layer, respectively. The base layer carries the overall luminance distribution, and its mapping deviation reflects the degree of preservation of large-scale light and dark structures. The detail layer carries local texture, and its mapping deviation reflects the degree of preservation of edge and texture information. The deviations of the two layers are added together according to preset weights to form the total value of the tone fidelity term. The smaller the value of the tone fidelity term, the higher the fidelity of the tone mapping process for the original luminance information.

[0082] Next, a gamut adaptation term is constructed. The gamut adaptation term measures the deviation between the chromaticity components after gamut mapping and the gamut boundary of the target digital printing device. Its function is to constrain gamut mapping to accurately map colors outside the printing gamut into the target gamut. The gamut adaptation term is calculated as follows: for each pixel in the image, the minimum chromaticity distance between the mapped chromaticity value and the gamut boundary of the printing device is calculated. If the mapped chromaticity value falls within the printing gamut, the gamut adaptation term for that pixel is zero, indicating no further adjustment is needed. If the mapped chromaticity value exceeds the printing gamut, the gamut adaptation term for that pixel is the square of the Euclidean distance between the mapped chromaticity value and the nearest point on the gamut boundary. The gamut adaptation terms of each pixel are summed over the entire image to form the total value of the gamut adaptation term. The smaller the value of the gamut adaptation term, the more pixels the chromaticity values ​​successfully adapt to the color reproduction capability of the printing device.

[0083] Next, coupling constraints are constructed. These constraints establish a connection between the tone mapping parameter vector and the gamut mapping parameter vector, coordinating the optimization process of the two types of parameters using the weight map generated in step S200. The coupling constraints are calculated as follows: for each pixel in the image, the square of the difference between the local mapping gradient of the tone mapping parameter and the local mapping gradient of the gamut mapping parameter at that pixel location is calculated, and then multiplied by the weight value of that pixel in the weight map. The local mapping gradient reflects the rate of change of the parameter in the pixel's neighborhood. The smaller the absolute value of the difference between the gradients of the tone mapping parameter and the gamut mapping parameter, the more consistent the adjustment direction and magnitude of the two mappings at that pixel location. The weight values ​​embed the coupling strength distribution in the weight map into the coupling constraints: at pixel locations with larger weight values, the gradient difference is amplified and penalized, increasing the contribution of the coupling constraints at that location. This drives the optimization process to impose stronger consistency constraints on the adjustment pace of the two types of parameters at that location, making the tone mapping and gamut mapping highly coordinated in the detail region. For pixels with smaller weights, the gradient difference is penalized, the contribution of the coupling constraint term at that location is reduced, and the color gamut mapping parameter gains more independent adjustment space at that location. The total value of the coupling constraint term is obtained by summing the coupling constraint terms of each pixel over the entire image.

[0084] Finally, the tone fidelity term, gamut adaptation term, and coupling constraint term are weighted and summed according to preset proportional coefficients to construct a complete joint objective function. The joint objective function uses the tone mapping parameter vector and gamut mapping parameter vector to be optimized as independent variables, and the weighted sum of the three terms as the dependent variable. The tone fidelity term and gamut adaptation term ensure the processing quality of luminance compression and gamut mapping, respectively, while the coupling constraint term establishes a pixel-level adaptive coupling relationship between the two using a weighted graph. The smaller the value of the joint objective function, the better the overall performance in terms of luminance fidelity, gamut matching, and joint coordination.

[0085] Step S300 provided in this embodiment of the invention further includes:

[0086] Introducing spectral constraints into the color gamut adaptation item specifically includes: pre-generating a spectral reflectance model of the target digital printing device, replacing the chromaticity distance metric in the color gamut adaptation item with a spectral reflectance difference metric, so that the solution of the color gamut mapping parameter vector simultaneously minimizes the reflectance difference of the color at each wavelength before and after mapping.

[0087] The specific implementation method is as follows:

[0088] The color gamut fitting term constructed in the above steps is based on the chromaticity distance metric, which measures the degree of color gamut fitting by calculating the Euclidean distance between the mapped chromaticity values ​​and the color gamut boundary of the printing equipment. This chromaticity distance-based metric only considers the geometric deviation of colors in the three dimensions of the CIELAB space, without addressing differences in the reflectance characteristics of colors at the physical spectral level. When two colors satisfy the chromaticity distance constraint in the CIELAB space, the human eye cannot perceive the difference between them under a specific light source. However, the spectral reflectance curves of the two colors may differ under different light sources, a phenomenon known as metamerism. Metamerism refers to the phenomenon where two colors appear the same under one light source but exhibit different colors under another. The fundamental reason is that although the chromaticity coordinates of the two colors match, their spectral reflectance distributions at different wavelengths are not consistent.

[0089] The purpose of introducing spectral constraints is to further improve the color accuracy of gamut mapping from the physical dimension of spectral reflectance. Before constructing the gamut adaptation terms, a spectral reflectance model of the target digital printing equipment is pre-generated. The spectral reflectance model describes the reflectance distribution of each color output by the printing equipment at various wavelengths under a specific paper and ink combination. The spectral reflectance model is established by using a spectrophotometer to measure the color targets printed on typical paper by the target digital printing equipment with different dot area ratios of each primary color ink, obtaining the spectral reflectance values ​​of each color patch at various wavelengths in the visible light wavelength range of 380 nm to 730 nm. The device input values ​​of each color patch, i.e., the dot area ratio or ink volume value of each primary color channel, are mapped to the corresponding spectral reflectance data, and a spectral reflectance model covering the device input value space is generated through an interpolation algorithm. The spectral reflectance model takes the ink volume value of each channel of the device as input and the full-band spectral reflectance curve of the corresponding color as output.

[0090] The chromaticity distance metric in the original gamut adaptation terms is replaced with a spectral reflectance difference metric. The replacement is as follows: for each channel's ink quantity value after gamut mapping, the spectral reflectance curve corresponding to that ink quantity value combination is predicted using a spectral reflectance model; for the ideal spectral reflectance curve corresponding to the original chromaticity values ​​before mapping, it is estimated using the original RGB data of the source high dynamic range image via a spectral reconstruction algorithm. The square of the difference in reflectance values ​​at each wavelength between the two spectral reflectance curves before and after mapping is calculated, and summed over all wavelengths to obtain the spectral reflectance difference metric for that pixel. The spectral reflectance difference metrics for each pixel are summed across the entire image to form the total value of the gamut adaptation terms after introducing spectral constraints.

[0091] By introducing spectral constraints, the solution for the gamut mapping parameter vector no longer focuses solely on optimizing the compressed chromaticity distance, but simultaneously minimizes the difference in reflectance of colors at various wavelengths before and after mapping. By shifting from a chromaticity distance-based metric to a spectral reflectance-based metric, the benchmark for color matching changes from "looking the same to the human eye under a specific light source" to "having the same physical spectral reflectance distribution." This fundamentally eliminates the impact of metamerism on color reproduction accuracy, ensuring that printed materials are highly consistent with the original artwork under any light source.

[0092] The following technical effects were achieved through this step:

[0093] First, this step incorporates tone mapping and gamut mapping into a unified optimization framework by constructing a joint objective function that includes tone fidelity, gamut adaptation, and coupling constraints. This ends the situation in traditional serial processing where the two are optimized independently and lack coordination. In the joint objective function, the three terms are defined independently yet cooperate with each other. The tone fidelity and gamut adaptation terms ensure the quality of their respective processing, while the coupling constraints establish a quantitative correlation between the two types of parameters using a weight graph. Through the unified optimization of the joint objective function, the tone mapping parameters and gamut mapping parameters guide each other and converge collaboratively in each iteration based on the current optimal intermediate results.

[0094] Secondly, this step introduces spectral constraints into the color gamut adaptation section, upgrading the benchmark for color matching from chromaticity distance to the difference in reflectance across the entire spectral band. This ensures that the optimization of color gamut mapping not only considers visual matching under specific light sources but also strives for consistency in the distribution of source image colors and printed reproduction colors along the physical spectral dimension. By replacing traditional chromaticity distance metrics with a measure based on spectral reflectance differences, the impact of metamerism on the accuracy of printed colors is fundamentally eliminated.

[0095] S400: The joint objective function is solved by an iterative optimization algorithm. Each iteration includes: alternately optimizing the tone mapping parameter vector and the gamut mapping parameter vector, and making feedback corrections to the two types of parameters based on the weight graph.

[0096] The above steps construct a joint objective function that includes a tone fidelity term, a gamut adaptation term, and a coupling constraint term, incorporating tone mapping and gamut mapping into a unified optimization framework. The joint objective function uses the tone mapping parameter vector and the gamut mapping parameter vector to be optimized as independent variables; a smaller function value indicates better overall performance across the three levels of brightness fidelity, gamut matching, and joint coordination. Due to the presence of the coupling constraint term in the joint objective function, the optimization of tone mapping parameters and gamut mapping parameters is not independent: each adjustment to the tone mapping parameters will affect the direction of the optimal solution for the gamut mapping parameters through the coupling constraint term, and vice versa. If a joint solution method is used to optimize all variables simultaneously, the convergence speed will become slow when the parameter vector dimension is high, and the solution process will be highly sensitive to the choice of initial values.

[0097] Step S400 provided in this invention includes:

[0098] Determining whether the iteration termination condition is met includes:

[0099] When the decrease in the joint objective function between two consecutive iterations is less than a preset decrease threshold, the iteration termination condition is determined to be met.

[0100] Alternatively, when the number of iterations reaches the preset maximum number of iterations, the iteration termination condition is determined to be met.

[0101] Alternatively, when the change magnitude of both the tone mapping parameter vector and the gamut mapping parameter vector in two adjacent iterations is less than a preset change threshold, the iteration termination condition is determined to be met.

[0102] The specific implementation method is as follows:

[0103] First, the tone mapping parameter vector and gamut mapping parameter vector are initialized. The tone mapping parameter vector takes the default parameter values ​​for logarithmic domain linear mapping. This set of parameter values ​​linearly maps the maximum brightness value of the source high dynamic range image to the maximum brightness value that the printing equipment can reproduce, and the minimum brightness value linearly maps to the minimum brightness value that the printing equipment can reproduce. The gamut mapping parameter vector takes the default parameter values ​​for nearest-point gamut clipping. This set of parameter values ​​maps colors outside the printing gamut to the nearest point on the gamut boundary along a direction that maintains hue in the same way.

[0104] After entering the iterative loop, two optimization sub-steps are executed in each iteration. The first optimization sub-step fixes the gamut mapping parameter vector to its current value and optimizes the hue mapping parameter vector. With the gamut mapping parameter vector fixed, the gamut adaptation term in the joint objective function is a constant and does not participate in this round of optimization. The parts of the hue fidelity term and the coupling constraint term related to the hue mapping parameters constitute the optimization sub-objective function. The gradient difference between the hue mapping parameter and the gamut mapping parameter in the coupling constraint term is simplified to a regularization term that depends only on the gradient of the hue mapping parameter because the gamut mapping parameter is fixed in this round. The gradient of this optimization sub-objective function with respect to the hue mapping parameter is calculated, and the hue mapping parameter vector is updated along the gradient descent direction.

[0105] The second optimization sub-step fixes the tone mapping parameter vector to its updated value and optimizes the gamut mapping parameter vector. With the tone mapping parameter vector fixed, the tone fidelity term in the joint objective function is a constant and does not participate in this round of optimization. The parts of the gamut adaptation term and coupling constraint term related to the gamut mapping parameters constitute the optimization sub-objective function. The gradient difference between the tone mapping parameters and the gamut mapping parameters in the coupling constraint term is simplified to a regularization term that depends only on the gradient of the gamut mapping parameters because the tone mapping parameters are fixed in this round. The gradient of this optimization sub-objective function with respect to the gamut mapping parameters is calculated, and the gamut mapping parameter vector is updated along the gradient descent direction.

[0106] After each iteration, it is determined whether the iteration termination condition is met. There are three iteration termination conditions; the iteration terminates when any one of them is met. The first termination condition is that the decrease in the joint objective function between two consecutive iterations is less than a preset decrease threshold. The decrease threshold is determined based on the magnitude of each cost term in the joint objective function. The joint objective function value is the weighted sum of the hue fidelity term, gamut adaptation term, and coupling constraint term. The normalized values ​​of each term usually fall between zero and one. The decrease threshold is preset to a dimensionless minimum positive value. When the decrease in the joint objective function value between two consecutive iterations is less than this threshold, it indicates that the joint objective function has approached its minimum value in the current search direction, and further iteration is unlikely to bring further substantial improvement; the optimization has saturated.

[0107] The second termination condition is reaching the preset maximum number of iterations. The maximum number of iterations is determined comprehensively based on the image resolution and real-time requirements. The purpose of the maximum number of iterations is to set an upper limit on the computational time of the optimization process, preventing the iteration process from continuing indefinitely when convergence is slow. In digital printing image color separation processing scenarios, the total processing time from source image input to color separation result output is limited by the printing production cycle. If the upper limit of the number of iterations is too small, the optimization may not converge sufficiently, and the tone mapping parameters and gamut mapping parameters may remain far from the joint optimal solution; if the upper limit is too large, the processing time will exceed the allowable waiting time for the printing job. Therefore, the maximum number of iterations is determined through offline pre-experiments based on the pixel scale of typical HDR images and the timeliness requirements of the printing job. For example, several representative HDR images are selected, and the convergence rate and convergence time of the joint objective function under different numbers of iterations are statistically analyzed. The number of iterations that enables approximately 95% of the sample images to converge is selected as the preset maximum number of iterations.

[0108] The third termination condition is that the changes in both the tone mapping parameter vector and the gamut mapping parameter vector in two consecutive iterations are less than a preset change threshold. The change threshold is determined based on preliminary experiments. Several representative HDR images are selected, and the decay trend of the parameter vector changes during iteration is statistically analyzed. The change threshold corresponding to the inflection point where the decay curve enters a stable phase is selected as the preset change threshold. When the changes in both types of parameter vectors between adjacent iterations are less than their respective change thresholds, it indicates that the parameters have stabilized near their optimal values, and further iterations will not cause effective updates to the parameter values.

[0109] If none of the above three conditions are met, the next iteration continues. Before the next iteration begins, the updated weight map is regenerated based on the intermediate luminance and intermediate chrominance components of the current round, as described in step S200. The updated weight map is then used to calculate the coupling constraint terms in the two optimization sub-steps of the next iteration. When the iteration termination condition is met, the tone mapping parameter vector obtained in the last iteration is output as the optimal tone mapping parameter vector, and the gamut mapping parameter vector obtained in the last iteration is output as the optimal gamut mapping parameter vector.

[0110] Step S400 provided in this embodiment of the invention further includes:

[0111] Before performing the iterative optimization algorithm, the source high dynamic range image is decomposed into low-resolution and high-resolution images.

[0112] First, the joint objective function is solved on the low-resolution image to obtain the global optimization parameter range;

[0113] Then, on the high-resolution image, the global optimization parameter vector is used as the initial value or constraint boundary to perform local fine optimization.

[0114] The specific implementation method is as follows:

[0115] Before starting the iterative optimization algorithm, a multi-scale decomposition preprocessing is performed on the source high dynamic range image. The purpose of multi-scale decomposition is to reduce computational complexity while providing reasonable initial values ​​and constraint boundaries for fine-grained optimization at high resolution.

[0116] First, the original RGB color data of the source high dynamic range image loaded in step S100 is downsampled. The downsampling uses a Gaussian pyramid downsampling method, performing multiple convolutions to smooth the original image and then sampling every other row and column to generate a medium-resolution image at half resolution and a low-resolution image at one-quarter resolution. The total number of pixels in the downsampled low-resolution image is approximately one-quarter of the original image, and the medium-resolution image is approximately half of the original image.

[0117] Then, on a low-resolution image at one-quarter resolution, the color space conversion and hierarchical decomposition in step S100 and the weight map generation in step S200 are performed. The joint objective function constructed in step S300 is used as the optimization objective, and the alternating iterative optimization algorithm described in the previous steps is employed to solve the problem. Since the number of pixels in the low-resolution image is far less than that in the original image, the computational cost of each summation term in the joint objective function is significantly reduced, allowing the iterative solution process to be completed quickly. The global optimization parameter vector obtained at low resolution reflects the optimal joint mapping relationship between the overall brightness and color distribution of the scene.

[0118] Next, using the globally optimized parameter vector obtained at low resolution as the initial value, a second round of iterative solving is performed on a medium-resolution image at half resolution. Using the globally optimized parameter vector as the starting point for this round, the parameters have already undergone preliminary optimization at low resolution, and their values ​​are close to the region of the global optimum. The number of iterations in this round is significantly less than in the case starting from the default initial values. After solving, the globally optimized parameter vector at medium resolution is obtained.

[0119] Finally, on the original high-resolution image, the global optimization parameter vector obtained at medium resolution is used as the initial value for iterative optimization, and a preset neighborhood interval of this parameter vector is used as the constraint boundary to perform local fine optimization. The constraint boundary limits the search range of the parameters at the original high resolution to within the neighborhood interval, ensuring that the fine optimization at high resolution is performed in a local region near the global optimum. This utilizes the global convergence direction determined at low resolution and further adjusts the parameters within the local range to adapt to the fine textures and details in the high-resolution image. The parameter values ​​at low resolution have already fixed the adjustment scheme for the global illumination and color distribution. Under this constraint, the fine optimization at high resolution only searches for local fine adjustments to each parameter, ensuring the stability of the iterative solution at each resolution level.

[0120] The following technical effects were achieved through this step:

[0121] First, this step employs an alternating optimization strategy to iteratively solve the joint objective function. In each iteration, one parameter vector is fixed while the other is optimized. The updated values ​​of the tone mapping parameters are used to optimize the gamut mapping parameters in this round, and the feedback response of the gamut mapping parameters is realized within this round. The two types of parameters are updated alternately within the framework of the joint objective function, rather than being fixed separately after being executed serially. This avoids the problem of cumulative distortion caused by the independent optimization of mapping parameters in traditional serial processing.

[0122] Second, before iterative optimization, this step performs multi-scale decomposition preprocessing on the source image. It quickly solves for the global optimization parameter vector on the low-resolution image, and then performs local fine-tuning optimization within the constraint boundaries on the high-resolution image, using the global optimization result as the initial value. By employing a layer-by-layer optimization strategy from low to high resolution, the global optimization parameters obtained at low resolution serve as a reliable starting point for high-resolution optimization. Local fine-tuning at high resolution only requires fine-tuning near the global optimum, avoiding the high computational cost and convergence uncertainty faced by optimizing directly from scratch at the original resolution.

[0123] S500: Apply the optimal tone mapping parameter vector and the optimal gamut mapping parameter vector obtained when the iteration termination condition is met to the source high dynamic range image to obtain the multi-channel ink volume value after color separation and output it.

[0124] The above steps iteratively solved the joint objective function using an alternating optimization strategy, obtaining the optimal tone mapping parameter vector and the optimal gamut mapping parameter vector. The optimal tone mapping parameter vector defines the mapping relationship from the original luminance component of the source high dynamic range image to the reproducible luminance range of the printing equipment, while the optimal gamut mapping parameter vector defines the mapping relationship from the original chrominance component to the reproducible chrominance values ​​within the color gamut of the printing equipment. These two parameter vectors are currently in numerical form and have not yet been applied to the source image to produce actual pixel value transformations.

[0125] Step S500 provided in this embodiment of the invention further includes:

[0126] The multi-channel ink volume values ​​include at least the cyan channel ink volume value, the magenta channel ink volume value, the yellow channel ink volume value, and the black channel ink volume value;

[0127] When the target digital printing equipment includes a light-colored ink channel, the multi-channel ink volume value also includes a light cyan channel ink volume value and a light magenta channel ink volume value.

[0128] When the target digital printing equipment includes spot color ink channels, the multi-channel ink volume value also includes the spot color channel ink volume value.

[0129] The specific implementation method is as follows:

[0130] The optimal tone mapping parameter vector obtained in step S400 is applied to the original luminance component L of each pixel obtained in step S100. The tone mapping parameter vector defines a transformation curve from the original luminance to the mapped luminance. For each pixel, its original luminance value is mapped to a mapped luminance value along this transformation curve. The dynamic range of the original luminance component is compressed into the luminance range reproducible by the printing equipment through tone mapping.

[0131] The optimal gamut mapping parameter vector obtained in step S400 is applied to the original chromaticity components a and b of each pixel obtained in step S100. The gamut mapping parameter vector defines the three-dimensional spatial transformation relationship from the original chromaticity value to the mapped chromaticity value. For each pixel, its original chromaticity component is mapped to the mapped chromaticity component along this transformation relationship. Colors in the original chromaticity components that exceed the gamut boundary of the printing equipment are adapted to the gamut through gamut mapping.

[0132] After completing the above two steps, each pixel possesses one mapped luminance component and two mapped chrominance components, namely the L, a, and b values ​​after hue mapping and gamut mapping. The mapped L, a, and b values ​​are then converted to the color space of the target digital printing equipment. The color space conversion method is as follows: first, the mapped L, a, and b values ​​are converted to CIEXYZ tristimulus values ​​using the CIE-specified inverse conversion formula; then, the XYZ tristimulus values ​​are converted to channel values ​​of the device-dependent color space using the target digital printing equipment's color profile file. The target digital printing equipment's color profile file is a standardized file describing the equipment's color reproduction characteristics, typically using the ICC Profile format specified by the International Color Consortium. The color profile file establishes a lookup table mapping relationship between the device-dependent color space and the device-independent CIELAB space, enabling mutual conversion between the two spaces. The mapped channel values ​​are linearized and converted into multi-channel ink volume values ​​after color separation. The ink volume value of each channel is expressed as a percentage or dot area ratio, ranging from zero to 100%, where zero indicates no ink supply to the channel and 100% indicates full ink supply to the channel.

[0133] Multi-channel ink volume values ​​include at least the ink volume values ​​of four standard four-color channels: cyan, magenta, yellow, and black. These four-color ink volume values ​​cover the basic color reproduction requirements of most digital printing equipment.

[0134] When the target digital printing equipment includes a light-color ink channel, the multi-channel ink volume values ​​also include light cyan and light magenta channel ink volume values. The introduction of the light-color ink channel aims to replace standard four-color inks in highlight areas, reducing ink droplet graininess and improving image smoothness and detail. The light-color ink channel ink volume value is calculated based on the standard four-color channel ink volume values ​​and the highlight area ink distribution strategy. When the target digital printing equipment includes a spot color ink channel, the multi-channel ink volume values ​​also include the spot color channel ink volume value. Spot color inks are pre-mixed, specific color inks used to print special colors that standard four-color inks cannot accurately reproduce, such as gold, silver, or corporate brand colors. The spot color channel ink volume value is calculated based on the chromaticity information of the corresponding spot color area in the source image, combined with the spectral characteristics data of the spot color ink.

[0135] The multi-channel ink volume values ​​after color separation are packaged and output according to the data interface format of the printing equipment control system. The output data contains a two-dimensional matrix of ink volume values ​​for each channel. The width and height of the matrix are consistent with the resolution of the source image, and each element is the ink volume value of the corresponding channel for that pixel. The output color separation data can be directly sent to the imaging engine of the digital printing equipment for plate making and printing operations.

[0136] The following technical effects were achieved through this step:

[0137] First, this step independently applies the optimal tone mapping parameter vector and the optimal gamut mapping parameter vector obtained through joint optimization to the luminance and chrominance components, respectively. During the optimization process, both parameter vectors have been optimized and adjusted to coordinate with the other parameter vector through the coupling constraint term of the joint objective function. During application, they are mapped in their respective dimensions according to the coordinated direction. Compared to the serial processing mode where the tone mapping parameter adjustment does not consider subsequent gamut mapping requirements and the gamut mapping parameter adjustment does not consider the previous luminance compression level, the application of the jointly optimized parameter combination ensures that the luminance and chrominance mappings present a coordinated visual effect in the final image.

[0138] Secondly, this step expands the output of the light-color ink channel and the spot color ink channel according to the actual configuration of the target digital printing equipment when outputting multi-channel ink volume values. Compared with the color separation method that only outputs the standard four-color ink volume values, expanding the light-color ink output for equipment containing the light-color ink channel improves the tonal performance of the highlight area, and expanding the spot color ink output for equipment containing the spot color ink channel to accurately reproduce special colors, so that the color separation method can match the color reproduction capabilities of digital printing equipment with different configurations.

[0139] Example 2, as Figure 4 As shown, based on the same inventive concept as the color separation processing method for digital printing images based on tone mapping provided in Embodiment 1, this embodiment of the invention also provides a color separation processing system for digital printing images based on tone mapping, the system comprising:

[0140] The color space conversion module 11 is used to load the original RGB color data of the source high dynamic range image and convert the original RGB color data to CIELAB space to obtain the original luminance component and original chrominance component of each pixel.

[0141] The weight map generation module 12 is used to calculate local contrast parameters based on the original luminance components, calculate saturation parameters based on the original chrominance components, and generate a weight map based on the local contrast parameters and the saturation parameters. The weight map is used to characterize the coupling strength between tone mapping and gamut mapping at each pixel position.

[0142] The joint objective function construction module 13 is used to construct a joint objective function, which includes at least a hue fidelity term, a gamut adaptation term, and a coupling constraint term.

[0143] The iterative optimization solution module 14 is used to solve the joint objective function through an iterative optimization algorithm. Each iteration includes: alternately optimizing the tone mapping parameter vector and the gamut mapping parameter vector, and performing feedback correction on the two types of parameters based on the weight graph.

[0144] The color separation output module 15 is used to apply the optimal tone mapping parameter vector and the optimal color gamut mapping parameter vector obtained when the iteration termination condition is met to the source high dynamic range image, obtain the multi-channel ink volume value after color separation and output it.

[0145] In one embodiment, the color space conversion module 11 is further configured to perform layer decomposition on the original luminance components to obtain a base layer representing the overall luminance distribution and a detail layer representing local texture details;

[0146] The joint objective function also includes a detail retention term, which is constructed based on the differential compression constraints between the base layer and the detail layer, and is used to characterize the balance between overall brightness compression and local detail retention during tone mapping.

[0147] In one embodiment, the weight graph generation module 12 is further configured to generate a weight graph, including:

[0148] Based on the local contrast parameter and the saturation parameter, the normalized values ​​of the local contrast parameter and the normalized values ​​of the saturation parameter are weighted and summed according to a preset weighting coefficient to calculate the weight value of each pixel.

[0149] Wherein, the normalized value of the local contrast parameter is the ratio of the local contrast parameter to the sum of the maximum and minimum local contrast parameters in the entire image, and the normalized value of the saturation parameter is the ratio of the saturation parameter to the sum of the maximum and minimum saturation parameters in the entire image.

[0150] Feedback correction is performed on the tone mapping parameter vector and the gamut mapping parameter vector based on the weight map, including:

[0151] For pixel positions in the weighted graph where the weight value is greater than a preset high weight threshold, the coefficient of the coupling constraint term is increased in the next iteration;

[0152] For pixel positions in the weighted graph where the weight value is less than a preset low weight threshold, the coefficient of the coupling constraint term is reduced in the next iteration.

[0153] The weight graph is updated synchronously in each iteration, and the weight graph is replaced with an updated weight graph recalculated based on the intermediate results obtained from the current iteration.

[0154] The weight map is updated after each iteration based on the intermediate optimization results of the current iteration. Specifically, based on the intermediate luminance and intermediate chrominance components obtained in this iteration, the local contrast and saturation parameters of each pixel are recalculated, and a new weight map is generated for the next iteration.

[0155] In one embodiment, the joint objective function construction module 13 is further configured to introduce spectral constraints into the color gamut adaptation term, specifically including: pre-generating a spectral reflectance model of the target digital printing device, replacing the chromaticity distance metric in the color gamut adaptation term with a spectral reflectance difference metric, so that the solution of the color gamut mapping parameter vector simultaneously minimizes the reflectance difference of the color before and after mapping at each wavelength.

[0156] In one embodiment, the iterative optimization solution module 14 is further configured to determine whether the iteration termination condition is met, including:

[0157] When the decrease in the joint objective function between two consecutive iterations is less than a preset decrease threshold, the iteration termination condition is determined to be met.

[0158] Alternatively, when the number of iterations reaches the preset maximum number of iterations, the iteration termination condition is determined to be met.

[0159] Alternatively, when the change magnitude of both the tone mapping parameter vector and the gamut mapping parameter vector in two adjacent iterations is less than a preset change threshold, the iteration termination condition is determined to be met.

[0160] Before performing the iterative optimization algorithm, the source high dynamic range image is decomposed into low-resolution and high-resolution images.

[0161] First, the joint objective function is solved on the low-resolution image to obtain the global optimization parameter range;

[0162] Then, on the high-resolution image, the global optimization parameter vector is used as the initial value or constraint boundary to perform local fine optimization.

[0163] In one embodiment, the color separation output module 15 is further configured to obtain multi-channel ink volume values ​​after color separation, including:

[0164] The multi-channel ink volume values ​​include at least the cyan channel ink volume value, the magenta channel ink volume value, the yellow channel ink volume value, and the black channel ink volume value;

[0165] When the target digital printing equipment includes a light-colored ink channel, the multi-channel ink volume value also includes a light cyan channel ink volume value and a light magenta channel ink volume value.

[0166] When the target digital printing equipment includes spot color ink channels, the multi-channel ink volume value also includes the spot color channel ink volume value.

Claims

1. A method for color separation of a digital print image based on tone mapping, characterized in that, include: Load the original RGB color data of the source high dynamic range image, convert the original RGB color data to CIELAB space, and obtain the original luminance component and original chrominance component of each pixel; Local contrast parameters are calculated based on the original luminance components, saturation parameters are calculated based on the original chrominance components, and a weight map is generated based on the local contrast parameters and the saturation parameters. The weight map is used to characterize the coupling strength between tone mapping and gamut mapping at each pixel position. Construct a joint objective function, which includes at least a hue fidelity term, a gamut adaptation term, and a coupling constraint term; The joint objective function is solved by an iterative optimization algorithm. Each iteration includes: alternately optimizing the tone mapping parameter vector and the gamut mapping parameter vector, and making feedback corrections to the two types of parameters based on the weight graph. The optimal tone mapping parameter vector and the optimal gamut mapping parameter vector obtained when the iteration termination condition is met are applied to the source high dynamic range image to obtain the multi-channel ink volume values ​​after color separation and output them.

2. The tone mapping based digital printing image separation process method according to claim 1, characterized in that, Also includes: Perform layer decomposition on the original luminance components to obtain a base layer representing the overall luminance distribution and a detail layer representing local texture details; The joint objective function also includes a detail retention term, which is constructed based on the differential compression constraints between the base layer and the detail layer, and is used to characterize the balance between overall brightness compression and local detail retention during tone mapping.

3. The tone mapping based digital printing image separation process method of claim 1, wherein, Generate a weighted graph, including: Based on the local contrast parameter and the saturation parameter, the normalized values ​​of the local contrast parameter and the normalized values ​​of the saturation parameter are weighted and summed according to a preset weighting coefficient to calculate the weight value of each pixel. Wherein, the normalized value of the local contrast parameter is the ratio of the local contrast parameter to the sum of the maximum and minimum local contrast parameters in the entire image, and the normalized value of the saturation parameter is the ratio of the saturation parameter to the sum of the maximum and minimum saturation parameters in the entire image.

4. The tone mapping based digital printing image separation process method of claim 1, wherein, Feedback correction is performed on the tone mapping parameter vector and the gamut mapping parameter vector based on the weight map, including: For pixel positions in the weighted graph where the weight value is greater than a preset high weight threshold, the coefficient of the coupling constraint term is increased in the next iteration; For pixel positions in the weighted graph where the weight value is less than a preset low weight threshold, the coefficient of the coupling constraint term is reduced in the next iteration. The weight graph is updated synchronously in each iteration, and the weight graph is replaced with an updated weight graph recalculated based on the intermediate results obtained from the current iteration.

5. The tone mapping based digital printing image separation process method according to claim 4, wherein, The weight map is updated after each iteration based on the intermediate optimization results of the current iteration. Specifically, based on the intermediate luminance and intermediate chrominance components obtained in this iteration, the local contrast and saturation parameters of each pixel are recalculated, and a new weight map is generated for the next iteration.

6. The tone-mapping based digital printing image separation process method of claim 1, wherein, Determining whether the iteration termination condition is met includes: When the decrease in the joint objective function between two consecutive iterations is less than a preset decrease threshold, the iteration termination condition is determined to be met. Alternatively, when the number of iterations reaches the preset maximum number of iterations, the iteration termination condition is determined to be met. Alternatively, when the change magnitude of both the tone mapping parameter vector and the gamut mapping parameter vector in two adjacent iterations is less than a preset change threshold, the iteration termination condition is determined to be met.

7. The method of claim 1, wherein the method further comprises: Also includes: Before performing the iterative optimization algorithm, the source high dynamic range image is decomposed into low-resolution and high-resolution images. First, the joint objective function is solved on the low-resolution image to obtain the global optimization parameter range; Then, on the high-resolution image, the global optimization parameter vector is used as the initial value or constraint boundary to perform local fine optimization.

8. The method of claim 1, wherein the method further comprises: Obtain the multi-channel ink volume values ​​after color separation, including: The multi-channel ink volume values ​​include at least the cyan channel ink volume value, the magenta channel ink volume value, the yellow channel ink volume value, and the black channel ink volume value; When the target digital printing equipment includes a light-colored ink channel, the multi-channel ink volume value also includes a light cyan channel ink volume value and a light magenta channel ink volume value. When the target digital printing equipment includes spot color ink channels, the multi-channel ink volume value also includes the spot color channel ink volume value.

9. The method of claim 1, wherein the method further comprises: Also includes: Introducing spectral constraints into the color gamut adaptation item specifically includes: pre-generating a spectral reflectance model of the target digital printing device, replacing the chromaticity distance metric in the color gamut adaptation item with a spectral reflectance difference metric, so that the solution of the color gamut mapping parameter vector simultaneously minimizes the reflectance difference of the color at each wavelength before and after mapping.

10. A system for color separation of a digitally printed image based on tone mapping, characterized in that, The system for implementing the method according to any one of claims 1 to 9, the system comprising: The color space conversion module is used to load the original RGB color data of the source high dynamic range image and convert the original RGB color data to CIELAB space to obtain the original luminance component and original chrominance component of each pixel. The weight map generation module is used to calculate local contrast parameters based on the original luminance components, calculate saturation parameters based on the original chrominance components, and generate a weight map based on the local contrast parameters and the saturation parameters. The weight map is used to characterize the coupling strength between tone mapping and gamut mapping at each pixel position. A joint objective function construction module is used to construct a joint objective function, which includes at least a hue fidelity term, a gamut adaptation term, and a coupling constraint term. The iterative optimization solution module is used to solve the joint objective function through an iterative optimization algorithm. Each iteration includes: alternately optimizing the tone mapping parameter vector and the gamut mapping parameter vector, and performing feedback correction on the two types of parameters based on the weight graph. The color separation output module is used to apply the optimal tone mapping parameter vector and the optimal color gamut mapping parameter vector obtained when the iteration termination condition is met to the source high dynamic range image, obtain the multi-channel ink volume value after color separation and output it.

Images