Quantitative identification method of laser printing file seal ink time sequence based on LBP-color difference analysis

Abstract

The invention relates to an LBP-chromatic aberration analysis-based quantitative identification method for a laser printing file hippeak time sequence, which comprises the following steps of: performing image acquisition and preprocessing on a plurality of laser printing files (according to the sequential time sequence of file printing and stamping) with seals in a manner of'sequentially printing after cinnabar 'and'sequentially printing after cinnabar'; and obtaining two groups of ink surface images with uniform resolution, which are divided into'one-after-one 'and'one-after-one-one', and establishing a double-index identification model based on color conspicuousness degree and distribution uniformity degree. A double-index identification model based on the color obvious degree and the color distribution uniformity degree is established in advance, scientific and systematic comparison identification is conducted on the cross coverage time sequence of the hippeastrum in the laser printing file, and a stable and quantifiable judgment basis is provided for identification work in the field. The innovative method effectively breaks through the high dependence of a traditional inspection technology on subjective experience, and obviously improves the objectivity, accuracy and repeatability of an identification result.

Description

Technical Field

[0001] This invention relates to the field of ink timing determination technology in laser-printed documents, specifically a quantitative identification method for ink timing in laser-printed documents based on LBP-color difference analysis. Background Art

[0002] The examination of the ink sequence in laser-printed documents has always been an important research topic in the field of document examination. Currently, the methods for this type of examination are mainly divided into two categories: traditional physical evidence analysis and physicochemical testing analysis.

[0003] Traditional evidence analysis focuses on the overlapping relationship between laser-printed text and seal ink, emphasizing the continuity of the text and the distribution of the ink, often employing methods such as stereomicroscopy and fluorescence examination. While these methods offer some visual appeal, they heavily rely on the experience and judgment of the forensic personnel, making it difficult to ensure accuracy in complex cases.

[0004] In physicochemical testing and analysis, Raman spectroscopy is often used to analyze the components of printing ink. Thin-layer chromatography, which integrates extraction and development, is used to study the classification of photosensitive printing ink. However, this method has its drawbacks. The Raman spectrometer and thin-layer chromatograph are expensive and difficult for grassroots testing institutions to equip, which limits the widespread application of the method. In addition, the equipment needs to be calibrated and maintained regularly, and the consumption of consumables (such as chromatographic plates and solvents) is large, resulting in high long-term operating costs. Summary of the Invention

[0005] (a) Technical problems to be solved

[0006] This invention provides a quantitative identification method for the ink sequence of laser-printed documents based on LBP-color difference analysis, which solves the problems of traditional physical evidence analysis methods being highly subjective and lacking quantitative standards, as well as the high cost of instruments required for traditional physicochemical testing and analysis methods, which makes them inconvenient for widespread application.

[0007] (2) Technical solution

[0008] To achieve the above objectives, the present invention provides the following technical solution: a method for quantitative identification of the ink sequence of laser-printed documents based on LBP-color difference analysis, comprising the following steps:

[0009] Step 1: By distinguishing multiple laser-printed documents with stamps into "red first then black" and "black first then red", images are acquired and preprocessed to obtain two sets of ink surface images with the same resolution, namely "red first then black" and "black first then red".

[0010] Step 2: Establish a dual-index recognition model based on color intensity and distribution uniformity for the two sets of ink surface images from Step 1;

[0011] Step 3: Obtain the ink surface image of the laser-printed document whose ink time sequence is to be analyzed, and quantify the color intensity and distribution uniformity of the ink surface image respectively;

[0012] Step 4: Compare the ink surface image obtained in Step 3 with the dual-index recognition model based on color intensity and color distribution uniformity in Step 2, thereby identifying the ink sequence of the laser-printed document in Step 3.

[0013] In a further preferred embodiment, the method for establishing the dual-index recognition model based on color intensity and distribution uniformity in step 2 is as follows:

[0014] Step 21: Quantify the color intensity of the ink distribution area and the pure black ink area in the ink surface image from Step 1 to obtain a recognition model based on color intensity.

[0015] The method for establishing a recognition model based on color intensity is as follows:

[0016] Step 211: By performing color difference quantification analysis on the ink distribution area and the pure black ink area in the ink surface images divided into "red first, then black" in Step 1, the color difference characteristics of a set of ink surface images divided into "red first, then black" in Step 1 are obtained.

[0017] Step 212: By performing color difference quantification analysis on the ink distribution area and the pure black ink area in the ink surface image divided into "ink first, vermilion later" in Step 1, the color difference characteristics of another set of ink surface images divided into "ink first, vermilion later" in Step 1 are obtained.

[0018] Step 213: Establish a recognition model based on the degree of color obviousness by quantifying the color difference features of the two groups of ink surface images divided into "red first then black" and "black first then red".

[0019] Step 22: Quantify the color distribution uniformity of the ink surface image divided into "first ink, then vermilion" in Step 1 to obtain a recognition model based on the color distribution uniformity.

[0020] In a further preferred embodiment, the color difference quantification analysis calculation in step 211 or step 212 is as follows:

[0021] Randomly select several color points in the pure black ink area of ​​the ink surface image, calculate the average color value, and denot it as Lab1. The brightness value of the color is L1, the chromaticity value of the red-green channel of the color is a1, and the chromaticity value of the blue-yellow channel of the color is b1.

[0022] Randomly select several color points in the ink distribution area of ​​the ink surface image, calculate the average color value, and denot it as Lab2, where the brightness value of the color is L2, the chromaticity value of the red-green channel of the color is a2, and the chromaticity value of the blue-yellow channel of the color is b2.

[0023] Applying the color difference formula Calculate the color difference value ΔE between Lab1 and Lab2;

[0024] The chromatic difference feature of a set of ink surface images is quantified as the set of chromatic difference values ​​ΔE for each ink surface image in the set of ink surface images.

[0025] In a further preferred embodiment, the quantitative calculation of the color distribution uniformity in step 22 is as follows:

[0026] The ink surface image is binarized, and then all the pixels constituting the ink surface image are divided into multiple 3×3 windows. The number of pixels in each 3×3 window whose neighboring pixel gray value is greater than the gray value of the center pixel is counted and this number is denoted as LBP'. The feature value of LBP' is between 0 and 8.

[0027] The sum of the number of pixels with LBP' feature values ​​of 1-7 in each of the above windows is used as the ratio ρ to the number of pixels with LBP' feature value of 8 in each of the above windows to indicate the uniformity of color distribution. In the formula, t is the eigenvalue identifier, and a t Let t be the total number of pixels with feature value t.

[0028] In a further preferred embodiment, step 1 is divided into "ink first, then vermilion" or "vermilion first, then ink" according to the order of document printing and stamping. In step 1, the images of each ink surface are acquired by taking microscopic pictures of the area where vermilion and ink intersect under uniform lighting conditions. During microscopic pictures, the background area of ​​the paper and the non-target ink dot area are covered. The preprocessing after the ink surface image acquisition includes grayscale normalization, cropping and extraction of the target area image, and image denoising.

[0029] (3) Beneficial effects

[0030] Compared with existing technologies, this invention provides a quantitative identification method for the ink sequence of laser-printed documents based on LBP-color difference analysis, which has the following beneficial effects:

[0031] In this invention, a dual-index recognition model based on color intensity and color distribution uniformity is pre-established to scientifically and systematically compare and identify the cross-over sequence of ink (stamp ink and printed text) in laser-printed documents, providing a stable and quantifiable basis for identification work in this field. This innovative method effectively overcomes the high dependence of traditional testing techniques on subjective experience, significantly improving the objectivity, accuracy, and repeatability of the identification results.

[0032] Traditional methods of determining the chronological order of red and black ink rely on visual observation of the intersecting areas, including color intensity and edge morphology. This approach is susceptible to influences from personal experience, visual fatigue, and ambient lighting. This invention introduces two quantifiable objective indicators: color intensity and color distribution uniformity. This transforms ambiguous visual perception into precise numerical comparisons, eliminating human interference and ensuring high consistency in judgments of the same sample by different personnel or laboratories. Furthermore, by combining color intensity and distribution uniformity, a complementary analytical system is formed. Attached Figure Description

[0033] Figure 1 This is a flowchart of a method for quantitatively identifying the ink timing sequence of laser-printed documents based on LBP-color difference analysis, according to the implementation plan. Detailed Implementation

[0034] The following will clearly and completely describe the technical solutions in the embodiments of the present invention in conjunction with the accompanying drawings. Obviously, the described embodiments are only part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without making creative efforts are within the scope of protection of the present invention.

[0035] Please see Figure 1 A quantitative identification method for the ink sequence of laser-printed documents based on LBP-color difference analysis includes the following steps:

[0036] Step 1: Multiple laser-printed documents bearing stamps (according to the order of printing and stamping) are categorized into "red first, then black" and "black first, then red" images. Image acquisition and preprocessing are then performed to obtain two sets of ink surface images with uniform resolution, one for "red first, then black" and the other for "black first, then red." In Step 1, each ink surface image is acquired by microscopic photography of the red and black intersection areas under uniform lighting conditions, while the background area of ​​the paper and non-target ink dots are masked during the photographing process. Subsequent preprocessing after obtaining the ink surface images may include grayscale normalization, cropping and extraction of the target area image, and image denoising.

[0037] Step 2: Establish a dual-index recognition model based on color intensity and distribution uniformity for the two sets of ink surface images from Step 1.

[0038] When establishing a dual-index recognition model based on color intensity and distribution uniformity, firstly, the color intensity of the ink distribution area and the pure black ink area in the ink surface image from step 1 is quantified to obtain a recognition model based on color intensity. Then, the color distribution uniformity of the ink surface image divided into "ink first, vermilion second" in step 1 is quantified to obtain a recognition model based on color distribution uniformity.

[0039] The method for establishing a recognition model based on color intensity is as follows:

[0040] By performing color difference quantification analysis on the ink distribution area and the pure black ink area in the ink surface images divided into "vermilion first, then ink" in step 1, the color difference characteristics of a set of ink surface images divided into "vermilion first, then ink" in step 1 are obtained; by performing color difference quantification analysis on the ink distribution area and the pure black ink area in the ink surface images divided into "ink first, then vermilion" in step 1, the color difference characteristics of another set of ink surface images divided into "ink first, then vermilion" in step 1 are obtained.

[0041] Color difference quantization analysis can be performed by randomly selecting several color points from the pure black ink area in the ink image, calculating the average color value, and denoting it as Lab1. Here, the brightness value is L1, the red-green channel chromaticity value is a1, and the blue-yellow channel chromaticity value is b1. Similarly, several color points can be randomly selected from the ink distribution area in the ink image, calculating the average color value, and denoting it as Lab2. Here, the brightness value is L2, the red-green channel chromaticity value is a2, and the blue-yellow channel chromaticity value is b2. Then, the color difference formula is applied. Calculate the color difference value ΔE between Lab1 and Lab2. Quantify the color difference features of a set of ink surface images as the set of color difference values ​​ΔE for each ink surface image in the set. Then, establish a recognition model based on the degree of color obviousness by quantifying the differences in color difference features between two sets of ink surface images divided into "red first then black" and "black first then red".

[0042] It should be noted that this application uses the internationally recognized Lab color space to perform color analysis on the intersection of the laser-printed document and the stamp. The Lab color space is a color model based on human visual perception, where L represents the luminance value of a color, a represents the chromaticity value of the red-green channel, and b represents the chromaticity value of the blue-yellow channel. The Lab color space can objectively and digitally describe the visual differences in colors. In the Lab space, the color difference between two colors can be calculated using the aforementioned color difference formula (i.e., the CIE 1976 formula). The use of Lab color difference analysis, rather than absolute color values, is based on the following two considerations: First, in actual testing, the ink application and toner concentration are unstable, making it difficult to provide a unified standard for absolute color reference values; second, lighting conditions and environmental factors during image acquisition can lead to errors between the absolute color values ​​of the acquired images and the original physical colors. By strictly controlling the lighting environment, shooting distance, and instrument parameters during image acquisition, the relative stability of the obtained Lab color difference values ​​can be ensured, avoiding interference from fluctuations in absolute color values.

[0043] In addition, the method for establishing a recognition model based on the uniformity of color distribution (quantitative calculation of the uniformity of color distribution) is as follows:

[0044] The ink surface image, divided into "ink first, then vermilion," is binarized. Then, all pixels constituting the ink surface image are divided into multiple 3×3 windows. The number of pixels within each 3×3 window whose neighboring pixel grayscale value is greater than the center pixel's grayscale value is counted and denoted as LBP'. The feature value of LBP' is between 0 and 8. Next, the sum of the number of pixels with LBP' feature values ​​of 1-7 in each window is multiplied by the number of pixels with LBP' feature values ​​of 8 in each window, and the ratio ρ is used to indicate the uniformity of color distribution. In the formula, t is the eigenvalue identifier, and a t Let t be the total number of pixels with feature value t. Quantize the color distribution uniformity feature of a set of ink surface images divided into "ink first, then vermilion" as the set of each ρ value in the set of ink surface images.

[0045] It should be noted that the traditional LBP method uses a 3×3 pixel window as the basic unit, comparing the grayscale values ​​of the center pixel with its eight neighboring pixels. If a neighboring pixel's grayscale value is lower than the center pixel's, it is assigned a value of 0; otherwise, it is assigned a value of 1. Then, based on the binary sequence of the neighboring pixel values, an 8-bit binary number is obtained, which is then converted to decimal as the LBP texture feature value of the center pixel. The traditional LBP calculation formula is as follows:

[0046]

[0047] In the formula: P,R are the coordinates of the center pixel of the window, p represents the p-th pixel in the neighborhood, and gp g represents the grayscale value of the neighboring pixels. c Let x be the gray value of the center pixel, and sign(x) be the sign function. In traditional local binary mode, the feature value of a pixel contains the position information of its neighboring pixels, forming an 8-bit binary number as the feature value (when p=8).

[0048] However, in this application, the analysis of the uniformity of ink distribution does not require the specific location information of pixels in the traditional LBP method, but only needs to describe the local density of ink pixel distribution. Therefore, this application makes appropriate improvements to the LBP algorithm: only counting the number of pixels within a 3×3 window whose gray value is greater than that of the center pixel (i.e., the number of neighboring pixels assigned a value of 1), and using this number as the improved LBP feature value (denoted as LBP'). When the ink distribution is relatively uniform, the ink distribution (gray value) in the neighborhood around random pixels varies more, and the LBP' feature value is usually more dispersed between 1 and 8. Conversely, if the ink is concentrated in a specific area or unevenly distributed, the LBP' feature value will be more concentrated around 8.

[0049] Step 3: Obtain the ink surface image of the laser-printed document with the ink-color time sequence to be analyzed, and quantify the color intensity and distribution uniformity of the ink surface image. The acquisition and processing of the ink surface image of the laser-printed document with the ink-color time sequence to be analyzed are as in Step 1, and the quantification of its color intensity and distribution uniformity is as in Step 2.

[0050] Step 4: Compare the ink surface image obtained in Step 3 with the dual-index recognition model based on color intensity and color distribution uniformity in Step 2, thereby identifying the ink sequence of the laser-printed document in Step 3.

[0051] The following is the actual experimental process for establishing a dual-index recognition model based on color intensity and distribution uniformity in this application:

[0052] Experimental materials:

[0053] (1) Ink: To improve the practicality and representativeness of the experimental conclusions, based on the sales ranking of office consumables on the JD.com e-commerce platform, 10 popular and representative vermilion office inks were selected as experimental materials (as shown in Table 1). These materials cover different brands and different properties (oil-based, water-based, and photosensitive) to ensure the comprehensiveness and practical relevance of the experimental results.

[0054] Table 1. List of inks used in the experiment

[0055] Number Ink Brand Ink Model Ink Property Ink Description 1 Deli 9864 Oil-based "Quick-drying Oil-based" 2 Deli 9870 Oil-based "Fast-drying Oil-based" 3 Deli 9893 Water-based "Quick-drying Water-based" 4 MG (M&G) 97513 Oil-based None 5 MG (M&G) 97520 Oil-based None 6 MG (M&G) 97525 Water-based None 7 <![CDATA[ Simma (Cima)]]> 21532 Oil-based None 8 <![CDATA[ Shachihata (Domestic Flagship Brand) EHP-2 Oil-based None 9 <![CDATA[ Shachihata (Nissan Flagship) HGN-2 Oil-based None 10 MG (M&G) 97529 Oil-based Photosensitive Ink

[0056] (2) Seals: copper seals, wooden seals, ox horn seals, rubber seals and photosensitive seals.

[0057] (3) Paper: Select the top five office paper brands in terms of sales on JD.com (as shown in Table 2).

[0058] Table 2. List of Printing Papers Used in the Experiment

[0059] Number Brand Model Paper Standard Paper Size Paper Weight 1 Deli Coral Sea Q / NDL97 A4 <![CDATA[70gm 2 ]]> 2 M&G Q / APYVQAF4 A4 <![CDATA[70g / m 2 ]]> 3 Tianzhang Paper Industry Q / TZ-10 A4 <![CDATA[70g / m 2 ]]> 4 Mozi Culture Q / MZWH001-2023 A4 <![CDATA[70g / m 2 ]]> 5 APP Q / ASGD-1 A4 <![CDATA[70g / m 2 ]]>

[0060] Experimental equipment:

[0061] Canon MF5950dw laser printer, VHX-100F ultra-depth-of-field 3D microscope, Adobe Photoshop 2024 image processing software, and SPSS statistical software.

[0062] Experimental conditions and control of influencing factors:

[0063] The material of the stamp can affect the distribution of ink. Because different materials have varying adsorption capacities for ink, the diffusion or deposition characteristics of ink on the ink surface may differ. This experiment selected copper, wood, horn, rubber, and photosensitive stamps for stamping experiments. Due to the self-storing nature of photosensitive stamps, only No. 10 oil-based photosensitive ink was used for stamping. The other four types of stamps were dipped in different inks listed in Table 1 (Nos. 1 to 9) and stamped on laser-printed documents under the same stamping pressure, on the same paper, and with the same amount of ink.

[0064] After eliminating interference factors such as stamping pressure and paper material, no significant differences were found in the color intensity and uniformity of ink distribution between stamps of different materials. This indicates that stamp material is not a key factor affecting the microscopic distribution characteristics of ink and can be uniformly handled and variables simplified in the experiment. Therefore, to maintain experimental consistency and repeatability, rubber stamps were used for inks 1 to 9 in the formal experiment, and the same photosensitive stamp was used for ink 10 to reduce systematic errors caused by differences in stamp material.

[0065] Furthermore, the stamping pressure may affect the contact pressure between the ink and the ink mark surface, thereby altering the ink's diffusion range and deposition state. Before the formal experiment, 20 subjects of different genders and body types were selected. Using a copper stamp, subjects applied inks (numbers 1 to 9 listed in Table 1) at three pressure levels—light (10N±5N), medium (20N±5N), and heavy (30N±5N)—to laser-printed documents. Microscopic images of the experimental samples were collected, and the distribution characteristics under different pressure levels were analyzed. Significant differences were observed among the light, medium, and heavy pressure levels: at light pressure, the ink mark's color was less distinct and its distribution was more dispersed; at heavy pressure, the ink mark's color became more distinct, but due to the ink being squeezed and diffused, it appeared uniform; at medium pressure, the ink mark showed the best balance between color distinctness and uniformity. Therefore, the formal experiment uniformly used a medium pressure (20N±5N) for stamping to obtain the optimal balance between color distinctness and distribution uniformity.

[0066] Furthermore, different types of paper vary in fiber density, surface smoothness, and ink absorption capacity, which may affect the deposition and diffusion characteristics of ink on the ink surface. Before the formal experiment, different paper types were tested. A rubber stamp was used with medium pressure (20N±5N) to apply ink samples (numbers 1 to 9 in Table 1) to laser-printed documents on papers (numbers 1 to 5 in Table 2), and microscopic images were collected for analysis. No significant differences were observed in the ink distribution on different paper types (numbers 1 to 5). On rough paper surfaces, the outline clarity of the ink marks was slightly reduced, but the overall distribution pattern was similar. Therefore, the paper material had a relatively small impact on the experimental results. For the formal experiment, the highest-selling "Deli Coral Sea" (model Q / NDL97) A4 paper (number 1) with stable fiber density and absorption capacity was used to ensure consistency in experimental conditions.

[0067] Finally, the amount of ink applied directly determines the thickness and saturation of the ink deposit on the ink mark surface, thus affecting the color clarity and uniformity. Before the formal experiment, to clarify the effect of the application amount on ink distribution, this study set three application amounts: low (one application), medium (two applications), and high (three applications). Under the same stamping force (20N±5N), a rubber stamp was used to apply ink samples (numbers 1-9) to laser-printed documents on A4 paper using "Deli Coral Sea" (model Q / NDL97) ink, and microscopic images were collected for analysis. With a low application amount (one application), the ink mark color and uniformity were poor; with a high application amount (three applications), ink diffusion occurred, the color was darker, and ink accumulation was severe; with a medium application amount (two applications), the ink color and uniformity were better. Therefore, the formal experiment uniformly adopted a medium application amount (two applications) to ensure a balance between color clarity and uniformity.

[0068] Formal Experiment:

[0069] Preparation of experimental samples for "ink first, then vermilion":

[0070] (1) Select A4 paper of the "Deli Coral Sea" model Q / NDL97 and use a Canon MF5950dw laser printer to create a laser print document;

[0071] (2) Let the paper with the printed text stand for 24 hours in a room temperature, light-proof, and dry environment;

[0072] (3) Using a rubber stamp or a photosensitive stamp (photosensitive ink only), dip the stamp into the different inks listed in Table 1. Each time, the amount of ink dipped should be controlled to be moderate (20N±5N pressure, dip the stamp twice). Stamp the stamp on the surface of the laser-printed paper with moderate pressure (20N±5N). Make 10 stamp samples for each ink and number them from 1 to 10.

[0073] Preparation of experimental samples for "vermilion first, then ink":

[0074] (1) Using a rubber stamp or a photosensitive stamp (photosensitive ink only), dip the stamp into the different inks listed in Table 1. Each time, the amount of ink dipped should be controlled to be moderate (20N±5N pressure, dipped twice). On a blank A4 paper of "Deli Coral Sea" model Q / NDL97, stamp with moderate pressure (20N±5N). Make 10 stamp samples for each ink, numbered from 11 to 20.

[0075] (2) Let the above-mentioned stamp sample paper stand for 24 hours in a room temperature, light-proof, and dry environment;

[0076] (3) Print the text on the paper with the above stamp sample using a Canon MF5950dw laser printer.

[0077] Microscopic image acquisition:

[0078] Using a VHX-100F ultra-depth-of-field 3D microscope, after the sample documents were stored at room temperature, protected from light, and dried for 30 days, the areas where red and black ink intersected were photographed. To avoid image acquisition bias, a uniform magnification of 500x and a resolution of 1600x1200 were set.

[0079] Under a microscope, images were taken at the intersection of ink and printing ink to obtain images of different intersection areas. Since the ink surface of the "vermilion first, ink later" sample was relatively stable with small color difference changes, one image was collected for each sample. There was a color difference between the yellow and green light reflected at the intersection of the ink surface of the "ink first, vermilion later" sample. To ensure the accuracy of the experiment, three images were collected from the bright, mid-tone, and dark areas of the yellow-green light area, according to the color tone change.

[0080] The acquired microscopic images are saved in JPEG format to ensure data integrity and that they are uncompressed.

[0081] Image preprocessing:

[0082] The acquired images were standardized using Photoshop: the color mode was switched to Lab color space; the Levels tool was used to normalize the grayscale to [0, 255] using the Lightness (L) channel, unifying the image color information and ensuring consistent image quality. Photoshop was then used to remove the paper background and non-target ink areas outside the intersection areas, and the target area was cropped to complete the denoising and cropping process. Color values ​​for the ink and stain areas were extracted separately for accurate color analysis.

[0083] Quantitative analysis of color distinctness:

[0084] The steps for quantitatively analyzing the color distinctness of the overlapping areas in documents printed using the "ink first, then vermilion" method are as follows:

[0085] (1) There is no yellow-green ink reflection area on the "Ink First, Vermilion Later" document. Use the built-in color sampler in Photoshop, set the sampling size to "5x5 average", randomly select 30 color points, extract the Lab color value, and calculate its average value as Lab1(L1, a1, b1).

[0086] (2) There is a yellow-green ink reflection area on the above document. Using the built-in color sampler in Photoshop, set the sampling size to "5x5 average", randomly select 30 color points, extract the Lab color value, and calculate its average value as Lab2(L2, a2, b2).

[0087] (3) Using the aforementioned color difference formula (i.e., the CIE1976 color difference formula), calculate the color difference ΔE between Lab1 and Lab2:

[0088] (4) Record the color difference value ΔE for each sample as a quantitative indicator of color clarity, and denoted as data group ΔE1.

[0089] The steps for quantitatively analyzing the color distinctness of the overlapping areas in a "red first, then black" document are as follows:

[0090] (1) In the area where the color is darker in the "red first, black later" file, use the built-in color sampler in Photoshop, set the sampling size to "5x5 average", randomly select 30 color points, extract the Lab color value, and calculate its average value as Lab3(L3, a3, b3).

[0091] (2) In the areas of the above document where the color is lighter, use the built-in color sampler in Photoshop, set the sampling size to "5x5 average", randomly select 30 color points, extract the Lab color value, and calculate its average value as Lab4(L4, a4, b4).

[0092] (3) Using the CIE1976 color difference formula, calculate the color difference ΔE between Lab3 and Lab4:

[0093] (4) Record the color difference value ΔE for each sample as a quantitative indicator of color clarity, denoted as data set ΔE2.

[0094] Quantitative analysis of distribution uniformity:

[0095] The "ink first, then vermilion" document image was binarized according to color values. The LBP algorithm was used to calculate the LBP' feature value of each pixel in the image, and the distribution number of pixels with each LBP' feature value was counted. Simultaneously, the ratio of the sum of the number of pixels with feature values ​​of 1-7 to the number of pixels with a feature value of 8 was calculated to obtain the distribution uniformity coefficient ρ. The "ink first, then vermilion" samples exhibit significant color distribution non-uniformity, mainly manifested in the intersection areas of ink and printing ink. Due to the differences in permeability, adsorption, and diffusion between ink and printing ink, the color intersection areas exhibit complex microstructural features. This complex intersection structure results in a significant gradient change in color distribution in space, thus affecting the image uniformity. In this case, the uniformity can be evaluated by calculating the LBP' feature value of each pixel, counting the number of pixels with different feature values, and further using the feature value ratio. Furthermore, the "vermilion first, then ink" samples show relatively stable color distribution. Due to the strong covering power and diffusion of printing ink, the subsequent ink coverage forms a uniform coating on the surface, resulting in fewer transition areas between colors and less noticeable color differences. Therefore, quantitative analysis of uniformity is insufficient to reveal more complex structural features for the "red first, black later" sample, and the practical significance of quantitative analysis is limited. Therefore, quantitative analysis of the distribution uniformity of this group of samples is not performed.

[0096] Data statistics:

[0097] The ΔE and ρ values ​​of different ink samples on two types of documents were statistically analyzed. The mean, standard deviation and coefficient of variation were calculated respectively. SPSS statistical software was used to analyze the significant differences of ΔE and ρ values ​​between different types of inks on the two types of documents using one-way ANOVA.

[0098] Then, through regression analysis, the correlation between ΔE and ρ values ​​and the properties of printing oil was explored, and a statistical model was established.

[0099] The analysis of color distinctness is as follows:

[0100] The color difference ΔE1 value calculated using the color difference formula is shown in Table 3.

[0101] Table 3. Statistics on Color Differences of Various Printing Inks for the "Ink First, Vermilion Later" Style

[0102]

[0103] To further analyze the influence of ink properties on the ΔE1 value, a one-way ANOVA was performed using SPSS software. The results are shown in Table 4.

[0104] Table 4. Color difference ΔE and AN0VA analysis results for "Ink first, then Vermilion"

[0105]

[0106] The color difference ΔE2 for the "red first, then black" case was calculated using the same method, and the results are shown in Table 5.

[0107] Table 5. Color difference statistics at all overlapping sampling points of vermilion and black ink in the "vermilion first, then black" method.

[0108]

[0109] Using SPSS software, a one-way ANOVA was performed on the data in ΔE2, with the properties of the printing ink as a factor. The results are shown in Table 6.

[0110] Table 6. ANOVA Analysis Results of Color Difference ΔE2 for "Vermilion First, Black Later"

[0111]

[0112] Analysis of Tables 3, 4, 5, and 6 shows that, under the same stamping pressure, number of ink applications, and lighting conditions, the color difference of the ink surface in the "vermilion first, then ink" stamping method is not significantly correlated with the properties of the ink (significance > 0.05). Furthermore, the color difference ΔE of the "vermilion first, then ink" stamping method ranges from 1.82 to 5.36, which is smaller than the ΔE range (9.6-37.5) of the "ink first, then vermilion" method.

[0113] As shown in Tables 3 and 4, under the same stamping pressure, number of ink dips, and light conditions, the color difference of the ink surface of the "ink first, then vermilion" stamped text is not significantly related to the properties of the ink. The significance of all three sets of data is greater than 0.05.

[0114] Further analysis of the results in Table 1 reveals that the Deli9864 ink, which has the highest color difference level, is labeled as "instant-drying oil-based ink," suggesting that its composition may differ from other oil-based inks and may contain water-based quick-drying components, which could lead to abnormal color difference levels.

[0115] To avoid the influence of the data from Deli9864, after excluding the data for this model, a new analysis of variance (ANOVA) was performed on the remaining 9 data points. The results are shown in Table 7.

[0116] Table 7. ANOVA analysis results of color difference ΔE1 for "Ink first, then Vermilion" (excluding Deli9864).

[0117]

[0118] As shown in Table 7, after removing the Deli9864 data, the variance analysis of the three groups in ΔE1 showed significant differences (significance < 0.05) in both between-group and within-group analyses. This indicates a significant correlation between the color difference ΔE values ​​of different ink properties under the same stamping pressure, number of ink applications, and lighting conditions. In particular, water-based inks showed a more pronounced difference in yellow-green color reflection, with higher ΔE values, suggesting a fundamental difference in color reflection characteristics between water-based and oil-based inks.

[0119] The analysis of the uniformity of color distribution is as follows:

[0120] To quantify the uniformity of ink distribution under the "ink first, then vermilion" condition, the uniformity was characterized by the ratio of the total number of pixels with feature values ​​of 1-7 to the number of pixels with a feature value of 8, denoted as ρ. The results are shown in Table 8.

[0121] Table 8. Statistics on the uniformity of different types of printing ink

[0122]

[0123] To analyze the relationship between ink properties and uniformity, SPSS software was used to perform ANOVA analysis on the ρ value, and the results are shown in Table 9.

[0124] Table 9. Results of ANOVA analysis of uniformity of "ink first, then vermilion" (outliers removed)

[0125]

[0126] Among the different inks, Deli9864 exhibited the highest uniformity at 4.4, followed by Deli9870 at 3.2. Considering the differences in composition between Deli9864 and other oil-based inks, this may be related to its inclusion of water-based quick-drying components, leading to different color distribution patterns in the microstructure. Furthermore, Deli9870 had a relatively high ρ value, possibly related to the material and adsorption characteristics of the ink pad, resulting in lower ink saturation after printing compared to other inks, fewer yellow-green patches and color spots, and a relatively smaller number of pixels with a feature value of 8, thus increasing the ρ value.

[0127] To further avoid the influence of data from Deli9864 and Deli9870, after excluding these two sets of outlier data, ANOVA analysis of the ρ values ​​was performed using SPSS software. The results are shown in Table 10.

[0128] Table 10. Results of pANOVA analysis of uniformity of "ink first, then vermilion" (outliers removed)

[0129]

[0130] After removing the data from Deli9864 and Deli9870, the difference in uniformity ρ values ​​among different ink types reached a statistically significant level (significance < 0.05). This indicates that, under the same stamping pressure, number of ink dips, and lighting conditions, ink properties are significantly correlated with the uniformity of color distribution.

[0131] Further analysis revealed that water-based inks generally exhibit a higher degree of uniformity than oil-based inks. This is related to the stronger penetration and diffusion characteristics of water-based inks in paper fibers, resulting in a more uniform color coverage at the microscopic level.

[0132] Analysis of the causes of the difference between ΔE and ρ values:

[0133] In the "ink first, then vermilion" experiment, the ΔE and ρ values ​​were generally high, with ΔE values ​​typically ranging from 9.6 to 37.5 and ρ values ​​from 1.1 to 4.4. This is due to the characteristics of the toner layer in the laser-printed document. After heating and fixing, the laser-printed toner forms a rough surface with strong adsorption and weak diffusion, resulting in uneven distribution of ink on this surface. At this point, the ΔE and ρ values ​​fluctuate significantly depending on the type of ink. The ΔE value (average 26.89) and ρ value (average 2.1) of water-based inks are significantly higher than those of oil-based inks (average 18.18 and ρ value (average 1.28). This is because water-based inks typically use water and low-viscosity polar solvents such as ethylene glycol and propylene glycol as dispersion media, resulting in stronger ink flow and diffusion. Simultaneously, the surfactants and dispersants in water-based inks effectively reduce surface tension. [9] This process allows the ink to spread rapidly on the ink surface, and combined with its drying mechanism that relies on water evaporation, it ultimately forms a wide-ranging, thin, and relatively uniform pigment layer. This pigment layer has high smoothness, which is conducive to the orderly reflection of light, thus exhibiting higher ΔE and ρ values. In contrast, oil-based inks are mainly composed of long-chain hydrocarbons. [8] It has high viscosity, poor fluidity, and weak diffusion. Pigments tend to concentrate and deposit in local areas. Moreover, oily printing ink dries by forming an oxide film through air oxidation. After drying, the surface is uneven, which leads to enhanced light scattering, reduced color reflection efficiency, and relatively low ΔE and ρ values.

[0134] In the "vermilion first, ink later" experiment, the ΔE value was low and relatively stable, ranging from 1.82 to 5.36. This is mainly because when the laser printer prints on paper with pre-existing ink, the toner, through heating and fixing, firmly covers the ink surface, forming a uniform and complete coating, avoiding uneven color reflection. Under this sequence, due to the uniform toner coverage, the reflected light is more consistent, the ΔE value fluctuates less, and the ρ value is low and evenly distributed. The stable coating formed by the toner during heating and fixing reduces color difference variations caused by differences in ink properties, resulting in a low and stable ΔE value.

[0135] Experimental analysis results:

[0136] Through experimental statistical analysis, significant differences in color performance were observed on the surface of laser-printed ink marks due to different vermilion and ink timing sequences. Based on the ΔE value calculated using the color difference formula, the ΔE value for the "vermilion first, ink later" case ranged from 1.82 to 5.36, with a mean of approximately 3.14 and a standard deviation of 1.18, demonstrating a stable and consistent color difference level, and showing no significant correlation with the properties of the printing ink.

[0137] In contrast, when ink was applied first and then vermilion, the ΔE value fluctuated between 9.6 and 37.5, with a mean of approximately 21.5 and a standard deviation of 6.55, exhibiting a larger fluctuation range. One-way ANOVA revealed a statistically significant difference between the ΔE value and the ink properties.

[0138] Further analysis revealed that, under the "ink first, vermilion later" application method, Deli9864 had the highest ΔE value, with a color difference of 33.9, significantly higher than other inks. This suggests that Deli9864 may differ in composition from other oil-based inks. The label on Deli9864 indicates it is a "second-drying oil-based ink," suggesting it may contain water-based quick-drying components, thus exhibiting a higher level of color difference in its color reflection properties.

[0139] Even after removing Deli9864, the ΔE values ​​still showed significant differences among different inks. This indicates that the properties of the ink have a significant impact on color difference in the "ink first, vermilion later" application scenario, especially since the color difference value of water-based inks is generally higher than that of oil-based inks.

[0140] Therefore, the range and fluctuation of color difference values ​​can serve as important criteria for distinguishing the time sequence of red and black ink:

[0141] When the ΔE value is between 1.82 and 5.36 and the ρ value fluctuates slightly, it can be judged as "red first, then black".

[0142] When the ΔE value is between 9.6 and 37.5 and the ρ value fluctuates significantly, it can be judged as "ink first, then vermilion". In particular, when the ΔE value exceeds 25, it may be caused by water-based ink.

[0143] While embodiments of the present invention have been shown and described, it will be appreciated by those skilled in the art that various changes, modifications, substitutions, and variations may be made to these embodiments without departing from the principles and spirit of the invention, and that the scope of the invention is defined by the appended claims and their equivalents.

Claims

1. A quantitative identification method for the ink sequence of laser-printed documents based on LBP-color difference analysis. Includes, characterized in that, Includes the following steps: Step 1: By distinguishing multiple laser-printed documents with stamps into "red first then black" and "black first then red", images are acquired and preprocessed to obtain two sets of ink surface images with the same resolution, namely "red first then black" and "black first then red". Step 2: Establish a dual-index recognition model based on color intensity and distribution uniformity for the two sets of ink surface images from Step 1; Step 3: Obtain the ink surface image of the laser-printed document whose ink time sequence is to be analyzed, and quantify the color intensity and distribution uniformity of the ink surface image respectively; Step 4: Compare the ink surface image obtained in Step 3 with the dual-index recognition model based on color intensity and color distribution uniformity in Step 2, thereby identifying the ink sequence of the laser-printed document in Step 3.

2. The quantitative identification method for the ink sequence of laser-printed documents based on LBP-color difference analysis according to claim 1, characterized in that: The method for establishing the dual-index recognition model based on color intensity and distribution uniformity in step 2 is as follows: Step 21: Quantify the color intensity of the ink distribution area and the pure black ink area in the ink surface image from Step 1 to obtain a recognition model based on color intensity. Step 22: By quantifying the uniformity of color distribution in the ink surface image divided into "first ink then vermilion" in Step 1, a recognition model based on the uniformity of color distribution is obtained.

3. The quantitative identification method for the ink sequence of laser-printed documents based on LBP-color difference analysis according to claim 2, characterized in that: The method for establishing the color intensity-based recognition model in step 21 is as follows: Step 211: By performing color difference quantification analysis on the ink distribution area and the pure black ink area in the ink surface images divided into "red first, then black" in Step 1, the color difference characteristics of a set of ink surface images divided into "red first, then black" in Step 1 are obtained. Step 212: By performing color difference quantification analysis on the ink distribution area and the pure black ink area in the ink surface image divided into "ink first, vermilion later" in Step 1, the color difference characteristics of another set of ink surface images divided into "ink first, vermilion later" in Step 1 are obtained. Step 213: Establish a recognition model based on the degree of color obviousness by quantifying the color difference features of the two groups of ink surface images divided into "red first then black" and "black first then red".

4. The quantitative identification method for the ink sequence of laser-printed documents based on LBP-color difference analysis according to claim 3, characterized in that: The color difference quantification analysis calculation in step 211 or step 212 is as follows: Randomly select several color points in the pure black ink area of ​​the ink surface image, calculate the average color value, and denot it as Lab1. The brightness value of the color is L1, the chromaticity value of the red-green channel of the color is a1, and the chromaticity value of the blue-yellow channel of the color is b1. Randomly select several color points in the ink distribution area of ​​the ink surface image, calculate the average color value, and denot it as Lab2, where the brightness value of the color is L2, the chromaticity value of the red-green channel of the color is a2, and the chromaticity value of the blue-yellow channel of the color is b2. Applying the color difference formula Calculate the color difference ΔE between Lab1 and Lab2; The chromatic difference feature of a set of ink surface images is quantified as the set of chromatic difference values ​​ΔE for each ink surface image in the set of ink surface images.

5. The quantitative identification method for the ink sequence of laser-printed documents based on LBP-color difference analysis according to claim 2, characterized in that: The quantitative calculation of the uniformity of color distribution in step 22 is as follows: The ink surface image is binarized, and then all the pixels constituting the ink surface image are divided into multiple 3×3 windows. The number of pixels in each 3×3 window whose neighboring pixel gray value is greater than the gray value of the center pixel is counted and this number is denoted as LBP'. The feature value of LBP' is between 0 and 8. The sum of the number of pixels with LBP' feature values ​​of 1-7 in each of the above windows is used as the ratio ρ to the number of pixels with LBP' feature value of 8 in each of the above windows to indicate the uniformity of color distribution. In the formula, t is the eigenvalue identifier, and a t Let t be the total number of pixels with feature value t.

6. The method for quantitative identification of ink sequence in laser-printed documents based on LBP-color difference analysis according to claim 1, characterized in that: In step 1, the order of printing and stamping the document is divided into "ink first, then vermilion" or "vermilion first, then ink".

7. The quantitative identification method for the ink sequence of laser-printed documents based on LBP-color difference analysis according to claim 1, characterized in that: In step 1, the images of each ink surface were acquired by taking microscopic images of the areas where the red and black ink intersect under uniform lighting conditions.

8. The quantitative identification method for the ink sequence of laser-printed documents based on LBP-color difference analysis according to claim 7, characterized in that: In step 1, the background area of ​​the paper and the non-target ink dot area are masked when acquiring images of each ink surface.

9. A quantitative identification method for the ink sequence of laser-printed documents based on LBP-color difference analysis according to claim 7 or 8, characterized in that: The preprocessing of the ink surface image in step 1 includes grayscale normalization, cropping and extraction of the target region image, and image denoising.

Images