Information security control method and device when using a laser printer

Abstract

Embodiments of the present application disclose a method and device for information security control when using a laser printer. The method comprises: identifying text content of a first file using a recognition layer of a locally deployed AI model; determining a classified level of the first file to be printed according to the identified text content using a prediction layer of the locally deployed AI model; writing first data of the first file to a print cache of the laser printer and prohibiting writing print content to a hard disk of the laser printer when the classified level is a first classified level; printing the first file based on the first data stored in the print cache; after completing printing of the first file, writing preset data to the print cache to overwrite the first data in the print cache; after completing overwriting of the print cache, powering off a print part; and after powering off the print part, charging an ionizing rod.

Description

Technical Field

[0001] This invention relates to the field of laser technology, and in particular to an information security control method and apparatus for using a laser printer. Background Technology

[0002] Information security issues are becoming increasingly prominent during the daily use of laser printers. Currently, when laser printers process document printing tasks, printed information leaves residue on the printer, potentially leading to information leaks due to the inherent limitations of laser printer technology. In other words, laser printers in this technology pose significant information security risks, failing to meet the stringent security requirements for printing highly confidential documents and seriously threatening the information security of enterprises and organizations. Therefore, there is an urgent need for a method to effectively control the information security of laser printers to address these technical problems. Summary of the Invention

[0003] In view of this, embodiments of the present invention provide an information security control method and apparatus for using a laser printer. The technical solution of the present invention is implemented as follows:

[0004] A first aspect of this disclosure provides an information security control method for using a laser printer, the laser printer including a printing section and an ionization section; the printing section includes a toner cartridge; the ionization section includes an ionization bar arranged in parallel with the toner cartridge; the printing section and the ionization section are electrically isolated; the method includes: using a recognition layer of a locally deployed artificial intelligence (AI) model to recognize the text content of a first document; using a prediction layer of the locally deployed AI model to determine the security level of the first document to be printed based on the recognized text content; when the security level of the first document is a first security level, writing the first data of the first document into the print cache of the laser printer and prohibiting the writing of the first data of the first document into the hard disk of the laser printer; printing the first document based on the first data stored in the print cache; after the printing of the first document is completed, writing preset data into the print cache to overwrite the first data in the print cache; and after the print cache is overwritten, powering off the printing section;

[0005] After the printing section is powered off, the ionization rod is charged, causing the ionization rod to generate ions that neutralize the residual image on the drum; an isolation layer is provided on the side of the ionization rod away from the drum; the isolation layer is used to protect the circuitry inside the laser printer.

[0006] A second aspect of this disclosure provides an information security control device for using a laser printer, the laser printer including a printing section and an ionization section; the printing section includes a toner cartridge; the ionization section includes an ionization bar disposed parallel to the toner cartridge; the printing section and the ionization section are electrically isolated; the device includes: a determining module, configured to identify the text content of a first document using a recognition layer of a locally deployed AI model; and to determine the security level of the first document to be printed based on the identified text content using a prediction layer of the locally deployed AI model; and a writing module, configured to write the first data of the first document to the print buffer of the laser printer and prevent the printing of the first document when the security level of the first document is a first security level. The first data of the first file is written to the hard drive of the laser printer; the printing module is used to print the first file based on the first data stored in the printing cache; the overwrite module is used to write preset data to the printing cache to overwrite the first data in the printing cache after the printing of the first file is completed; the power-off module is used to power off the printing part after the printing cache is overwritten; the neutralization module is used to charge the ionization rod after the printing part is powered off, so that the ionization rod generates ions to neutralize the residual image on the drum; an isolation layer is provided on the side of the ionization rod away from the drum; the isolation layer is used to protect the circuitry inside the laser printer.

[0007] The technical solution provided in this disclosure provides end-to-end data leakage prevention, building a solid security defense for high-level classified documents. Local AI processing provides pre-emptive protection, using a locally deployed AI model to identify file content and determine its classification level, preventing file data from being uploaded to hard drives, the cloud, or external servers, thus reducing the risk of interception during transmission. Automatic classification level identification replaces manual labeling, reducing the risk of leakage due to human error or malicious tampering with classification levels. For first-level classified documents, data is only temporarily stored in the print cache and is prohibited from being written to the hard drive, avoiding potential data residue from long-term hard drive storage (such as extraction by recovery tools). Immediately after printing, the cache is overwritten with preset data, and a proactive erasure mechanism eliminates traces of temporary storage, preventing unauthorized reading of cached data. The ionization rod generates ions after power is turned off in the printing section, neutralizing residual charges and image traces formed on the drum surface due to electrostatic latent images, preventing the recovery of printed content using specialized equipment (such as the theft of drum residual images of classified documents), thus solving the pain point of hidden leakage of information by the physical carrier in traditional laser printing. Hardware isolation ensures effective data erasure. The printing and ionization sections are electrically isolated, and the ionization rod is equipped with an isolation layer to ensure that the high voltage or ion environment generated during ionization does not interfere with the printer's circuit safety. Simultaneously, it prevents residual electrical signals from the printing section from affecting the neutralization effect, achieving dual protection for both the erasure operation and device safety. Through AI-automated identification of security levels, dynamic cache management, and power outage / ionization linkage, the entire process of securely handling high-security documents requires no manual intervention. This reduces oversights caused by human error (such as forgetting to clear the cache or failing to process the toner cartridge in a timely manner) while maintaining the efficiency of the printing process. It is suitable for scenarios with extremely high information security requirements, such as confidential offices, medical records, and financial documents.

[0008] In summary, this method forms a closed-loop security control from data processing, storage, physical carrier to hardware protection, achieving full-cycle information security protection of preventing leakage before the event, controlling storage during the event, and clearing traces after the event, significantly improving the risk resistance of printing high-confidential documents. Attached Figure Description

[0009] The accompanying drawings, which are included to provide a further understanding of this application and form part of this application, illustrate exemplary embodiments and are used to explain this application, but do not constitute an undue limitation of this application. In the drawings:

[0010] Figure 1 A flowchart illustrating an information security control method for using a laser printer, provided in an embodiment of the present invention;

[0011] Figure 2 A flowchart illustrating an information security control method for using a laser printer, provided in an embodiment of the present invention;

[0012] Figure 3A schematic diagram of an information security control device for using a laser printer, provided in an embodiment of this disclosure;

[0013] Figure 4 This is a schematic diagram of the structure of an electronic device provided in an embodiment of the present disclosure. Detailed Implementation

[0014] To enable those skilled in the art to better understand the present invention, the technical solutions of the present invention will be clearly and completely described below with reference to the accompanying drawings of the embodiments of the present invention. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort should fall within the scope of protection of the present invention.

[0015] It should be noted that the terms "first," "second," etc., in the specification, claims, and accompanying drawings of this invention are used to distinguish similar objects and are not necessarily used to describe a specific order or sequence. It should be understood that such data can be interchanged where appropriate so that embodiments of the invention described herein can be implemented in orders other than those illustrated or described herein. Furthermore, the terms "including" and "having," and any variations thereof, are intended to cover non-exclusive inclusion; for example, a process, method, system, product, or apparatus that includes a series of steps or units is not necessarily limited to those explicitly listed, but may include other steps or units not explicitly listed or inherent to such processes, methods, products, or apparatus.

[0016] This disclosure discloses an information security control method for using a laser printer. The laser printer includes a printing section and an ionization section; the printing section includes a toner cartridge; the ionization section includes ionization rods arranged parallel to the toner cartridge; the printing section and the ionization section are electrically isolated.

[0017] The laser printer has a rectangular parallelepiped structure. For example, the outer dimensions of the rectangular casing are 450mm × 350mm × 280mm. Inside, the printing section and the ionization section are sequentially distributed along the paper travel path. The two sections are physically separated by an insulating partition (thickness greater than or equal to 3mm and less than or equal to 8mm, specifically 5mm) to achieve electrical isolation. For example, the insulation resistance of the insulating partition is ≥100MΩ.

[0018] In some embodiments, the printing section is located at the front of the printer (near the paper feed tray), and its core components include a toner cartridge (photosensitive drum), a laser (including a laser emitter, a beam splitter, and a first sensor), a developer, and a second sensor. The toner cartridge is horizontally mounted in the middle of the printing section, with its axis perpendicular to the paper travel direction. The distance between its surface and the developer's magnetic roller is 0.2 to 0.4 mm, for example, 0.3 mm, and its vertical distance from the laser's output port is 6 mm to 10 mm, for example, 8 mm, to ensure accurate laser projection.

[0019] Ionization section: Located behind the printing section (near the paper exit), parallel to the drum unit (ionization rod axis parallel to the drum axis, spacing 12mm to 17mm, e.g., 15mm). Core components include the ionization rod, high-voltage generator, and isolation layer. The ionization rod extends along the length of the drum unit, covering the entire effective printing area of ​​the drum unit (length 200mm to 400mm, e.g., 300mm, adaptable to A3 or A4 paper width).

[0020] In some embodiments, the distance between the ionization rod and the drum is (optimized range of 3-5mm); in some embodiments, the material composition of the isolation layer (ceramic-based composite insulating layer); exemplaryly, Table 1 below is an example of laser and / or parameters of a laser printer, and the materials and / or parameters in the table can be used individually or in combination in specific implementations.

[0021] part Structural parameters Material and Modular Properties Functional parameters toner cartridge 30mm in diameter, 300mm in length, and 5μm thick surface coating. The base material is aluminum alloy (elastic modulus 70 GPa), and the surface is coated with selenium alloy (hardness HV300). Operating voltage -600V (charging state), residual charge decay rate ≤5% / min Ionizing rod 5mm in diameter, 300mm in length, 10mm between needle tips <![CDATA[The main body is made of stainless steel (conductivity 1.5×10 7 S / m), and the tip is made of tungsten alloy (melting point 3422℃)]]> <![CDATA[Working voltage 5 - 8 kV (adjustable), ion generation rate ≥ 10 8 ions / s]]> isolation layer 2mm thick, covering the entire surface of the ionization rod on the side away from the drum. Polytetrafluoroethylene (insulation strength 20kV / mm, dielectric constant 2.1) Temperature resistance range: -20℃ to 260℃; aging resistance life: ≥5000 hours Insulating partition Thickness 5mm, area 400mm × 100mm <![CDATA[Glass fiber reinforced epoxy resin (insulation resistance ≥ 10¹ 0 Ω, flexural strength 150 MPa)]]> Withstand voltage ≥20kV to avoid electric field interference between the printed part and the ionized part.

[0022] In some embodiments, electrical isolation ensures dual safety: the printing section (exemplarily, operating voltage ≤ ±1kV) and the ionization section (e.g., high voltage: 5-8kV) are physically and electrically isolated by insulating partitions and isolation layers. This prevents the high voltage of the ionization rod from breaking down the circuit of the printing section (such as the laser driver module) and avoids the low-frequency electrical signals of the printing section from interfering with the uniformity of ion distribution, ensuring the stability of laser parameters (e.g., wavelength drift) and the reliability of ionization effect.

[0023] In some embodiments, the ionization rod precisely removes residual images. The ionization rod is positioned in parallel with the toner cartridge to specifically neutralize residual electrostatic images on the cartridge surface (especially the subtle charge distribution after printing high-density documents). Combined with feedback from a second sensor in the printing section (detecting paper printing quality), the ionization rod can dynamically adjust its operating voltage (e.g., increasing to 8kV when residual images are detected), ensuring residual image removal time is ≤2 seconds. This avoids the need for specialized equipment to restore printed content, making it suitable for scenarios involving confidential documents, medical records, etc.

[0024] In some limited embodiments, the selenium alloy coating on the surface of the drum has high hardness and can withstand long-term impact from ions generated by the ionization rod without aging; the tungsten alloy tip of the ionization rod has a high melting point, avoiding losses caused by corona discharge under high voltage; the polytetrafluoroethylene insulating layer has temperature resistance suitable for the internal working environment of the printer (50-60℃), ensuring stable insulation performance during long-term use.

[0025] In some embodiments, the optimization in conjunction with visual feedback may include: feedback data provided by the first sensor (monitoring the laser status) and the second sensor (monitoring the printing effect) of the printing section can be linked with the ionization section to adjust the working sequence (such as delaying the ionization for a specific time (0.5s) after printing to ensure accurate neutralization after the drum stops), which does not interfere with the normal printing process and ensures the thorough removal of afterimages, thereby achieving synergy between print quality optimization and information security protection.

[0026] In summary, through reasonable structural layout, parameter design, and material selection, this laser printer, while ensuring the function of adjusting printing parameters with visual feedback, achieves a dual improvement in print quality stability and information security reliability by leveraging the synergy between the ionization and printing components.

[0027] like Figure 1 As shown, the method includes:

[0028] S1110: Use the recognition layer of the locally deployed AI model to recognize the text content of the first document;

[0029] S1120: The prediction layer of the locally deployed AI model determines the security level of the first document to be printed based on the recognized text content;

[0030] S1130: When the security level of the first file is the first security level, write the first data of the first file into the print cache of the laser printer and prohibit writing the first data of the first file into the hard disk of the laser printer;

[0031] S1140: Print the first file based on the first data stored in the print cache;

[0032] S1150: After the first file is printed, preset data is written to the print buffer to overwrite the first data in the print buffer;

[0033] S1160: After the print buffer has been overwritten, power is turned off to the print section;

[0034] S1170: After the printing section is powered off, the ionization rod is charged, causing the ionization rod to generate ions that neutralize the residual image on the drum; an isolation layer is provided on the side of the ionization rod away from the drum; the isolation layer is used to protect the circuitry inside the laser printer.

[0035] This method is applied to the controller of a laser printer or laser printer.

[0036] In S1110-S1120, document recognition and security classification are performed. Specifically, this includes: a locally deployed AI model (based on the ResNet-50 architecture, recognition layer) recognizes the text content of the first document through Optical Character Recognition (OCR) and determines different security levels. The prediction layer combines keyword weights (weights of classified words ≥ 0.8) to determine the security level as the first security level (the highest level).

[0037] In the laser printer, the security level determination process relies on a locally deployed artificial intelligence (AI) model (integrating computer vision and natural language processing technologies), and is implemented through a three-level architecture of deep text feature analysis, multi-dimensional semantic association, and dynamic threshold adaptation.

[0038] S1110 may include at least one of the following:

[0039] Multimodal input adaptation: The recognition layer first receives the raw data from the first file (supporting multiple input formats such as scanned images, electronic documents (PDF / Word), and handwritten photos), and converts it into structured text through the following sub-steps:

[0040] Image preprocessing: For scanned documents or photos, an adaptive threshold segmentation algorithm is used to remove noise such as paper wrinkles and shadows (the noise removal rate is improved by 40% compared with traditional fixed threshold processing). Then, super-resolution reconstruction technology (based on generative adversarial network GAN) is used to upscale low-resolution text (such as resolution ≤150dpi) to 300dpi to ensure OCR recognition accuracy.

[0041] Multilingual hybrid recognition: Integrates a multilingual OCR engine (supporting 10 languages ​​including Chinese, English, Japanese, and Russian). For mixed texts of Chinese keywords and foreign language codes commonly found in classified documents (such as experimental data from the top-secret project X-73), it uses an attention mechanism to focus on cross-language entity boundaries, achieving a recognition accuracy of ≥99.2% (compared to ≤95% accuracy of traditional monolingual models).

[0042] Formatting feature preservation: During the text extraction process, layout information (such as bold font, header and footer, watermark) is preserved simultaneously. For example, if a red watermark of ★Top Secret★ is detected in the header (extracted through color channel separation technology), this formatting feature is used as a strong weighting factor for the security level determination (the weight value is preset to 0.8, which is higher than ordinary text keywords).

[0043] S1110 may also include structured text parsing, specifically including: word segmentation (based on a bidirectional LSTM model), part-of-speech tagging, and entity recognition (such as names of people, organizations, and project codes) of the extracted text, and constructing a three-level semantic tree of words, sentences, and paragraphs. For example, when identifying the text "2024 test launch parameters of a certain military region's XX missile," the system automatically labels the military region (organizational entity), the XX missile (classified project entity), and the test launch parameters (sensitive content), and associates the semantic dependency relationship among the three (the test launch parameters belong to the XX missile and are associated with a certain military region).

[0044] In some embodiments, S1120 may include: the prediction layer determines the security level based on the structured text features output by the recognition layer through a three-step method of feature weighting-semantic reasoning-threshold verification, and further improves the accuracy of the security level through dynamic semantic association and self-learning threshold adjustment.

[0045] Step 1: Weighted quantification of classified features. Construct a multi-level classified feature library, dividing text features into the following dimensions and assigning dynamic weights: Core keywords: explicit identifiers such as top secret, confidential, secret, classified, not disclosed, etc., with preset basic weights, for example, values ​​can be 0.6-0.9; Implicit sensitive entities: such as research project names, classified project codes (e.g., XX project Y satellite), sensitive data (e.g., coordinates, parameters, formulas), with weights of 0.4-0.7 (generated through real-time comparison with a classified entity library, which contains 100,000+ anonymized classified entity samples).

[0046] In some embodiments, the same keyword has different weights in different contexts. For example, this document is classified as secret (directly identified and weighted, for example, weighted at 0.8) vs. discussing the management methods of secret documents (indirectly mentioned and weighted, for example, weighted at 0.3). The contextual relevance is calculated by the BERT pre-trained model, and the weight values ​​are dynamically adjusted.

[0047] In some embodiments, format features such as header watermarks, bold fonts, and encryption identifiers (such as [encryption]) mentioned above have a weight of 0.3-0.5 (independent of the text content and used as an auxiliary criterion for judgment).

[0048] Example: For the core parameters (secret level) of a certain project of a certain classified unit in the text, the feature weighting result is: classified unit (0.5) + project (0.6) + core parameters (0.5) + secret level (0.8), and the weighted sum is 2.4.

[0049] The second step: semantic reasoning and security classification mapping, which may include: based on weighted summation, performing semantic reasoning through trained deep neural networks (DNNs) to output a security classification probability distribution (e.g., Level 1 (Top Secret): 15%, Level 2 (Confidential): 60%, Level 3 (Secret): 25%, Unclassified: 0%). The creativity lies in:

[0050] Cross-domain adaptation: The model integrates classified samples from multiple domains (military, government, finance, medical) through transfer learning to solve the problem of insufficient samples from a single domain. In cross-domain scenarios, the accuracy of classification is still ≥92% (the accuracy of traditional single-domain models is ≤80%).

[0051] Ambiguity resolution: For ambiguous keywords (such as "secret" being a common word in literary works), non-confidential scenarios are excluded by combining document topic classification (through LDA topic model). For example, if the document topic is novel writing, the weight of the word "secret" will automatically drop below 0.1.

[0052] Step 3: Dynamic threshold verification, which may include: preset basic thresholds (e.g., first security level ≥ 0.7, second security level ≥ 0.5, third security level ≥ 0.3), but dynamically adjusted by introducing a self-learning mechanism:

[0053] If a discrepancy is found in the classification of a certain type of document during five consecutive manual reviews (e.g., a military parameter document is actually classified as Class I but the model classifies it as Class II), the weight of the military parameter feature in that type of document will be automatically increased by 0.1 each time, and the corresponding threshold will be decreased (e.g., the Class I threshold will be reduced to 0.65).

[0054] Based on the usage scenarios of laser printers (determined by geographical location and user permissions), for example, in the intranet environment of a classified unit, the threshold is automatically tightened (the first level of security threshold is increased to 0.75) to reduce the risk of misjudgment.

[0055] In some embodiments, the AI ​​model runs entirely locally on the printer (relying on an embedded NPU chip with a computing power of 8 TOPS), avoiding the uploading of text data to the cloud and preventing the leakage of confidential information at the data link layer (traditional cloud-based judgment carries a transmission leakage risk of over 30%). When the final security level is determined to be Level 1, the laser printer's high-security process is triggered (data is only cached, writing to the hard drive is prohibited, and afterimages are cleared after printing, etc.), forming a closed-loop security control of identification, judgment, and protection, providing innovative intelligent protection for printing high-security documents.

[0056] The data storage and printing performed in S1130-S1140 may specifically include: the first data of the first file (such as a 10MB classified document) is written only to the 8GB print cache, and the controller prevents data from being written to the 256GB SSD hard drive through a hardware lock;

[0057] The laser projects a laser beam onto the toner cartridge based on the cached data, and the developer works in conjunction to complete the printing, maintaining a printing speed of 25 pages per minute (consistent with non-classified documents).

[0058] The S1150-S1170 processes perform cache overwriting and image clearing, which may include:

[0059] After printing is complete, preset random binary data (of the same size as the cache) is overwritten in a specified manner, for example, overwriting the cache at a speed of 3200Mbps, to ensure that the original data cannot be recovered.

[0060] In some embodiments, the overwrite can be completed within a specified time. The data to be overwritten can be all zeros, randomly generated data, or data from the next task that urgently needs to be printed, as long as the overwrite is completed within the specified time to erase the data stored in the cache.

[0061] After the buffer overwrite is complete, the printing section automatically shuts off power, cutting off the power supply to the toner cartridge and laser. Three seconds after the printing section is powered off, the high-voltage generator starts, and the ionization rod generates positive and negative ions under high voltage to neutralize residual images on the toner cartridge surface (residual image removal rate ≥99.9%). The isolation layer blocks high voltage interference to the printer's mainboard (operating voltage 12V). The high voltage can reach kilovolts.

[0062] In this embodiment, the local AI model avoids uploading files to the cloud. First-level confidential data is stored only in a volatile cache and is prohibited from being written to the hard drive, eliminating the risk of data theft or recovery at the storage layer. The ionization rod and drum are designed in parallel to specifically neutralize afterimages across the entire image. The charged particles generated at high speed under high voltage ensure the removal of residual charge on the drum surface in a very short time, solving the hidden leakage problem of traditional laser printer drums and reducing the risk of afterimages being further reduced by professional equipment. The printing section and ionization section are electrically isolated to prevent high voltage from damaging the printing circuit (such as the laser driver module) during the ionization process. The device's continuous operation stability requires no manual intervention from confidentiality level determination to afterimage removal. The secure processing time for a single first-level confidential document is short, meeting the needs of both security and efficiency in high-confidence scenarios.

[0063] In summary, this embodiment constructs a high-security document printing system that ensures data is not stored on disk, has zero cache residue, and completely removes afterimages through structural optimization, parameter matching, and process collaboration. It is suitable for confidential office scenarios with zero tolerance for information leakage.

[0064] In some embodiments, the recognition layer of the AI ​​model includes a first recognition unit, a prediction unit, and a second recognition unit connected in series.

[0065] like Figure 2As shown, the recognition layer using a locally deployed AI model to recognize the text content of the first document includes:

[0066] S1111: The first recognition unit processes the text content with a first scale parameter to obtain a first recognition result;

[0067] S1112: The prediction unit predicts the recognition direction based on the first recognition result;

[0068] S1113: The second recognition unit adjusts the first scale parameter according to the recognition direction to obtain the second scale parameter;

[0069] S1114: The first recognition unit processes the text content based on the second scale parameter to obtain a second recognition result, and fuses the first recognition result and the second recognition result to obtain a recognition result that matches the security level.

[0070] In some embodiments, the locally deployed AI model recognition layer is integrated into the embedded controller of the laser printer (equipped with an NPU neural network processing unit), employing a dual recognition unit + prediction unit collaborative architecture. The first recognition unit is built on a lightweight CNN (Convolutional Neural Network), focusing on rapidly extracting global text features and supporting large-scale text block processing. The second recognition unit is built on a Transformer attention mechanism, focusing on fine-grained feature parsing and supporting dynamic adjustment of recognition accuracy. The prediction unit embeds an LSTM (Long Short-Term Memory) network to analyze the uncertainty of the recognition results and predict optimization directions.

[0071] Taking the first document as an example, which is a scanned engineering document containing a classified project code, the text content includes information such as XX missile test data (2024-Q3) and component parameters: a certain engine thrust ≥120kN, etc. The identification process is as follows:

[0072] S1111: The first recognition unit processes the text content with the first scale parameter to obtain the first recognition result.

[0073] The first scale parameter definition: paragraph-level coarse-grained recognition parameters are adopted, specifically including: sliding window size (500 characters / window), convolution kernel size (7×7, focusing on global features), feature extraction stride (200 characters), and attention weight (assigning an initial weight of 0.3 to high-frequency words such as data parameters).

[0074] Processing procedure: The first recognition unit performs a global scan of the OCR text of the scanned document (which has been preprocessed to remove noise), extracts paragraph-level semantic features through a 7×7 convolution kernel, and outputs the first recognition result.

[0075] The test data of the XX missile identified suspected classified entities such as an engine, but due to the coarse scale, the 11 and II (Roman numerals) in the data were confused, and the numerical boundary of ≥120kN was blurred.

[0076] Output feature vector: contains 3 types of high-confidence entities (confidence 85%) and 2 types of low-confidence entities (a certain confidence level of 60% and a 120kN confidence level of 55%).

[0077] S1112: The prediction unit predicts the recognition direction based on the first recognition result, specifically including:

[0078] Uncertainty Analysis: The prediction unit analyzes the entropy value of the first recognition result through an LSTM network (the entropy value of low-confidence entities is >0.6, which is higher than the preset threshold of 0.4), and determines that there are two types of recognition ambiguity:

[0079] Character-level ambiguity: The number 11 in a certain text is confused with the Roman numeral II;

[0080] Boundary ambiguity: The numerical range of ≥120kN (whether 120 includes decimal places) is not clear.

[0081] Recognition direction prediction: Output optimization direction: Focus on the sentences containing 'a certain engine' and '≥120kN', improve character-level resolution, and refine numerical boundary recognition.

[0082] S1113: The second recognition unit adjusts the first scale parameter according to the recognition direction to obtain the second scale parameter.

[0083] Parameter adjustment logic: Based on the predicted recognition direction, the second recognition unit dynamically adjusts the coarse-grained global parameters to fine-grained local parameters.

[0084] Second scale parameters: the sliding window is reduced to sentence level (100 characters / window), the convolution kernel is switched to 1×1 (to improve character level resolution), the feature extraction stride is reduced to 50 characters, and the attention weights are tilted towards numbers, symbols, and item code suffixes (e.g., the 11kN weight is increased to 0.7).

[0085] S1114: The first recognition unit processes the text based on the second scale parameter, and the fusion result yields the final recognition result. The second recognition result output may include: the first recognition unit calls the second scale parameter to perform a local scan on the sentence containing an engine thrust ≥120kN: clearly distinguishing 11 as a number using a 1×1 convolution kernel (excluding Roman numeral II, as the context engine model usually uses Arabic numerals); accurately identifying the boundary of ≥120kN (120 is an integer, with no decimal places), increasing the confidence level to 92%.

[0086] Results fusion: A weighted fusion algorithm is adopted (the first identification result has a weight of 0.4, the second identification result has a weight of 0.6, focusing on fine-grained identification) to output the final identification results: clearly identify entities: XX missile (confidence level 90%), test data (confidence level 95%), a certain engine (confidence level 92%), thrust ≥120kN (confidence level 92%), providing high-precision feature basis for subsequent classification (such as the first classification level).

[0087] In some embodiments, dynamic scale adaptation to improve recognition accuracy may include: breaking through the limitations of traditional single-scale recognition, by using coarse-grained global scanning + fine-grained local focusing, the recognition accuracy of classified entities (such as project codes and sensitive values) is increased from 60%-70% of the first recognition result to more than 90% after fusion, especially solving the problems of character ambiguity and boundary blurring.

[0088] In some embodiments, computational resource optimization may include: using a first scale parameter (large window) to quickly locate suspected regions, avoiding indiscriminate fine-grained processing of the entire document, reducing redundant computation by more than 30% (such as over-parsing of non-confidential paragraphs), and adapting to the computational limitations of the laser printer's embedded controller.

[0089] In some embodiments, providing a reliable basis for security classification determination may include: high-confidence classified entity features in the fusion results (such as a certain engine ≥120kN) can be directly used as key inputs for security classification determination, thereby reducing the security classification determination error rate of the subsequent prediction layer to below 5% and significantly improving the reliability of high-security document recognition.

[0090] In some embodiments, this embodiment achieves a balance between efficiency and accuracy through dynamic adjustment of dual-scale parameters and result fusion, providing accurate pre-identification support for information security control of laser printers (such as cache isolation of high-security documents and removal of afterimages).

[0091] In some embodiments, the method further includes:

[0092] When the security level of the first file is the second security level, the first data of the first file is written to the print buffer of the laser printer and the writing of the first data of the first file to the hard disk of the laser printer is prohibited; the second security level is lower than the first security level.

[0093] After the first file is printed, preset data is written to the print cache to overwrite the first data in the print cache;

[0094] Keep the printing unit powered on.

[0095] In some embodiments, the second security level is a lower level of sensitivity than the first security level (such as confidential internal documents or draft business contracts, which need to prevent unauthorized access but do not require the highest level of physical image removal). For example, a mid-term product development report (second security level) from a technology company contains undisclosed technical parameters that are not core secrets. It is necessary to ensure that the data does not remain on the hard drive during printing, but the printing process can be kept running to improve subsequent printing efficiency.

[0096] In some embodiments, the print cache has a capacity of 8GB DDR4 DRAM (shared with the first security level, but the data allocation priority of the second security level is lower than that of the first security level), a read / write speed of 3200Mbps, and supports a single-overwrite algorithm (different from the three redundant overwrites of the first security level).

[0097] Hard drive: Hardware-level locking mechanism prohibits writing of second-level classified data (consistent with the first level of classified data), and only allows the storage of non-classified files (such as public document templates).

[0098] Printing section: The parameters of components such as laser, drum, and developer are the same as those in the first density scenario (e.g., drum diameter 32mm, laser wavelength 650nm). However, when the machine is powered on, the drum maintains low power standby (voltage -200V, lower than the working voltage of -600V) to avoid lifespan loss caused by frequent start-stop.

[0099] After the prediction layer of the local AI model determines that the product development mid-term report is classified as Level 2 confidentiality, the controller executes the following: write the first data (such as a 5MB document) to an encrypted partition of the 8GB print cache (independent of the non-confidential data partition, using AES-128 encryption); and lock the hard drive write interface through firmware-level instructions (continuously monitor hard drive I / O requests, and directly return permission denial for write instructions for Level 2 confidentiality data).

[0100] In some embodiments, the printing process and cache overwriting may include: completing printing based on the printed cache data (printing speed of 30 pages / minute, consistent with non-classified documents); after printing is completed, triggering the cache overwriting mechanism: preset data (randomly generated binary data stream) is written to the cache encrypted partition at a speed of 3200Mbps, with an overwriting time of ≤0.3s (since the second level of security does not require redundant overwriting, a single overwriting is sufficient to meet security requirements), and after overwriting, the cache data is restored to its initial blank state (verified by testing tools, the data recovery rate is ≤0.01%).

[0101] In some embodiments, the state of the printing section is maintained as follows: after the coverage is completed, the controller keeps the printing section powered on: the laser is in standby mode (power reduced to 10% of the working state), the drum maintains low voltage charging (-200V), the developer magnetic roller stops rotating but maintains toner adsorption; the ionization section is not started (because the second density level does not need to remove the drum residue, and keeping the printing section powered on reduces the warm-up time for the next print from 30 seconds to 5 seconds).

[0102] In summary, the second-level security processing flow retains core security measures such as preventing data from being stored on the hard drive and overwriting the cache (eliminating the risk of hard drive remnants). It also eliminates the need for power-off and ionization rod removal steps in the printing section, reducing the security processing time for a single document from 10 seconds for the first-level security to 3 seconds. This is suitable for high-frequency sensitive document printing scenarios within enterprises (such as non-core confidential documents shared between departments). Keeping the printing section powered on (low-power standby) reduces the number of start-stop cycles for the toner cartridge and laser (reducing start-stop cycles by more than 50 times per day), which, according to calculations, can extend toner cartridge life by 15% (from 30,000 pages to 34,500 pages), reducing equipment maintenance costs. The hierarchical processing logic dynamically adapts through a local AI model's security level determination module. When a second-level security document contains temporarily upgraded sensitive segments (such as inserted first-level security data), it can automatically trigger the first-level security process (such as overwriting the cache, power-off, and activating the ionization rod), achieving flexible security control with dynamic security level switching.

[0103] In summary, this embodiment designs a processing mechanism that maintains security without compromising efficiency for second-level confidentiality documents. Through hierarchical caching management and dynamic state maintenance, it ensures data integrity while balancing printing continuity and hardware economy in enterprise-level office scenarios.

[0104] In some embodiments, the method further includes:

[0105] When the security level of the first file is the third security level, the first data of the first file is written to the print buffer of the laser printer and the first data of the first file is allowed to be written to the hard disk of the laser printer; the third security level is lower than the second security level.

[0106] The first data stored in the print cache is overwritten with the second data from the second file to be printed.

[0107] In some instances, the third security level is a lower level of sensitivity than the second security level (such as documents that are publicly available within a company but whose external dissemination is restricted, such as departmental notices and meeting minutes). Its information sensitivity is relatively low, and strict physical image removal or power-off of the printed portion is not required, but data should be prevented from remaining on the hard drive or in the cache for extended periods. For example, a company's weekly departmental work summary (third security level) contains internal arrangements but non-core sensitive information, making a lightweight security control approach of temporary caching followed by subsequent file overwriting suitable.

[0108] In this disclosed embodiment, 8GB of DDR4 DRAM is reused (shared with the first and second security levels, with the third security level having the lowest data allocation priority), and a dynamic overwrite mechanism is supported (no need to preset random data, directly overwrite through subsequent file data).

[0109] In some embodiments, the hard drive is a 256 GB SSD, and the controller firmware restricts the writing of third-level security data (consistent with the first and second levels of security), allowing only the storage of non-security system files (such as drivers and log templates).

[0110] The storage and printing of Class 3 classified data may involve the following steps: The local AI model's prediction layer determines the classification as Class 3 through text recognition (e.g., a departmental work summary without classified keywords, with context indicating internal public access). The first data (a 3MB document) is written to the shared partition of the print cache (unencrypted due to low sensitivity). The controller monitors hard drive I / O requests in real time and disables write commands for Class 3 classified data (consistent with high-class data, to prevent accidental writes). The laser projects a laser onto the toner cartridge based on the cached data, and the developer completes the printing process without any additional security verification delays (unlike the multiple verifications of Class 1 classified data).

[0111] In some embodiments, the cache overwrite mechanism (dependent on subsequent files) may include: after printing is completed, instead of performing a preset data overwrite, the cached data is retained until the next print job: when the user prints a second file (such as next week's meeting agenda, non-classified or third-classified), the second data (2MB) is written to the cache shared partition, directly overwriting the storage address of the original third-classified data (cache physical address reuse, old data cannot be recovered); if there are no subsequent print jobs within 1 hour, the controller automatically triggers a low-priority overwrite (writing 1MB of blank data to overwrite the core area to avoid long-term idle residue).

[0112] In some embodiments, the printing section status control may include: keeping the printing section powered on throughout the process (consistent with the second security level), maintaining the drum unit at standby voltage (-200V), and ensuring that the laser is in a ready state to ensure that the next print does not require preheating (response time ≤ 3 seconds).

[0113] In some embodiments, the balance between lightweight security and efficiency may include: basic security measures for the third-level security process that prevent data from being stored on the hard drive (eliminating the risk of hard drive remnants), while reducing cache operation time by 30% (from 0.3 seconds for the second-level security to near zero) by naturally overwriting subsequent files instead of actively pre-setting data overwriting, adapting to the printing needs of high-frequency, low-sensitivity files (such as daily office documents for enterprises). The cache shared partition reuse and low-priority overwriting mechanism reduce memory usage (the average cache usage of third-level security data is ≤5%), avoiding the consumption of memory bandwidth by redundant overwriting in high-security scenarios, and improving the printer's multi-tasking capabilities (it can cache 3 copies of third-level security files + 1 copy of second-level security files simultaneously). As a supplement to the first and second levels of security, the third-level security process, through precise matching of sensitivity and security measures (such as unencrypted caching and natural overwriting), constructs a full-spectrum security architecture that provides heavy protection for high-security levels, efficiency for medium-security levels, and light control for low-security levels, meeting the information security needs of different scenarios and avoiding resource waste caused by excessive protection.

[0114] In summary, this embodiment designs a lightweight security process adapted to the sensitivity of third-level confidentiality documents, maximizing printing efficiency and resource utilization while ensuring data integrity, and improving the hierarchical security control system for laser printers.

[0115] In some embodiments, the method further includes:

[0116] When the security level is the third security level, the first data of the first file is written to the print cache and / or hard disk according to the current state of the laser printer.

[0117] In some embodiments, the third level of security is a lower level of sensitivity than the second level (such as publicly available internal enterprise process documents, departmental general reports, etc., which have lower information sensitivity and are allowed limited hard drive storage, but unnecessary residues must be avoided). Taking a manufacturing company's workshop daily production report (third level of security) as an example, it contains daily output data (not core confidential information), and the storage location needs to be flexibly selected according to the real-time status of the printer to balance security and printing efficiency.

[0118] In some embodiments, the print cache has a capacity of 8GB DDR 4DRAM, supports real-time utilization monitoring (accuracy 1%), and the idle threshold is set to 30% (i.e., the cache is considered idle when the remaining capacity is ≥2.4GB).

[0119] In some embodiments, the hard drive is configured with a temporary storage partition (50GB capacity, used only for third-level security files, using NTFS encryption format) and supports status monitoring (including storage space utilization and IO activity).

[0120] In some embodiments, status monitoring may include: integrating with the controller to collect the following parameters in real time:

[0121] Cache usage (e.g., currently using 40%, i.e., 3.2GB);

[0122] Free hard drive space (e.g., 30GB remaining);

[0123] Disk I / O activity level (≤30% is considered idle, >70% is considered busy).

[0124] In some embodiments, the determination of the third security level and the triggering of status monitoring may include: after the prediction layer of the local AI model determines the security level as the third security level through text recognition (such as the production daily report not containing classified keywords and historical print records being internal circulation), the controller starts the status monitoring module to obtain the current status data: Example 1: Cache utilization 20% (idle), hard disk free space 40GB, IO busyness 15% (idle); Example 2: Cache utilization 85% (stressful), hard disk free space 25GB, IO busyness 60% (moderate).

[0125] A state-based dynamic storage strategy may include: the controller selecting a storage path according to priority rules based on monitoring results. For example, the storage path for data at the third security level may be determined according to the following rules.

[0126] Rule 1: Prioritize cache storage (optimal efficiency): When cache utilization is ≤30% and disk I / O activity is >50%, only write the first piece of data (e.g., a 2MB report) to the unencrypted partition of the print cache (because the third security level has low sensitivity, encryption is not required, but it is independent of the system cache). Example 1: Executing this rule: 2MB of data is written to the cache, and the print response time is ≤1 second (no disk read / write latency).

[0127] Rule 2: Cache + Disk Co-location (Reliability First): When cache utilization is between 30% and 70% and the disk is idle, the first data is written to both the cache (for immediate printing) and the temporary disk partition (for backup in case of cache overflow). Disk storage uses a timestamp + automatic deletion mechanism (automatically cleared after 24 hours to avoid long-term residue). Example scenario: When printing a large 10MB report, 5MB is temporarily stored in the cache, and the remaining 5MB is written to the disk. During printing, the controller dynamically schedules data to ensure smooth printing.

[0128] Rule 3: Disk Storage Only (When Cache is Insufficient): When cache utilization is >70% (e.g., 85% in Example 2), the first data is written directly to a temporary disk partition. During printing, data is transferred in real-time via disk-cache stream (reading and printing simultaneously) to avoid print interruptions caused by cache overflow. Security: A third-level security tag is automatically associated with disk writes, prohibiting unauthorized users from accessing the partition through the printer interface. The data format is a dedicated printing stream (cannot be directly parsed into a document).

[0129] Cache data: After printing, it is retained until the next print job and is naturally overwritten by new data (same as the previous third-level security processing logic); Disk data: Deletion is triggered by conditions - if there is no need for secondary printing within 24 hours, it is automatically deleted; if there is a secondary printing, it is deleted immediately after use, and only the deletion log is retained (the log does not contain file content).

[0130] In this embodiment of the disclosure, resource utilization is dynamically optimized: the fixed strategy of prohibiting hard disk writing in the first and second security levels is broken, and storage is flexibly allocated according to the cache and hard disk status. In the third security level scenario, the cache utilization rate is increased by 25% (avoiding idle waste), the utilization rate of the hard disk temporary partition is controlled within 10% (preventing excessive occupation), and the overall printing efficiency is improved by 15% (reducing the waiting time caused by insufficient cache).

[0131] Balancing security and flexibility: Even if third-level classified data is written to the hard drive, it is protected by three layers of protection: encrypted partitions, automatic deletion, and special format, with a leakage risk of ≤0.1% (far lower than the 1% of non-classified files). At the same time, it meets the needs of low-sensitivity files for rapid batch processing (such as centralized printing of reports from multiple shifts in the workshop).

[0132] Intelligent adaptation with state awareness: The controller achieves storage strategy switching without manual intervention through real-time monitoring and rule engine, adapting to the printer's working status under different loads (such as automatically activating the hard drive when the cache is tight during peak hours, and prioritizing cache speed during idle hours), improving the device's adaptability in complex office scenarios.

[0133] In summary, this embodiment designs a state-driven dynamic storage mechanism for third-level security files, which maximizes the use of printer hardware resources while ensuring the security of basic information, and improves the flexibility and practicality of the hierarchical security system.

[0134] In some embodiments, the method further includes:

[0135] Before charging the ionization rod, the ionization window is opened to expose the ionization rod, which is arranged in parallel with the drum.

[0136] In this embodiment, the pre-opening trigger of the ionization window may include: when the first document's security level is the first security level, and after the print buffer overwriting and power-off of the printing section are completed (corresponding to the preceding step S1160), the controller initiates the ionization preparation process: sending an opening command to the stepper motor of the ionization window, the motor drives the ionization window to slide horizontally (at a speed of 5 mm / s), moving from the initial closed position (blocking the ionization rod) to the open position (fully exposing the ionization rod tip, with a stroke of 20 mm). The position sensor monitors in real time, and when the ionization window is fully open (blocking the infrared light path is released, and a high level is output), a signal indicating completion of opening is fed back to the controller (taking approximately 4 seconds).

[0137] The ionization rod charging and afterimage neutralization process may include: after the controller confirms that the ionization window is open, it triggers the high-voltage generator: the ionization rod generates positive and negative ions under high voltage, which are directionally projected onto the surface of the toner cartridge through the opened ionization window (at this time, the printing part is de-energized, the toner cartridge has no working voltage, and the afterimage charge is exposed).

[0138] The neutralization process lasts for a specified duration (the specified duration can be 3 seconds, etc., and can be dynamically adjusted according to the length of the selenium drum and the ion density) to ensure that the residual charge neutralization rate is ≥99.9% (through simulation calculation, the neutralization rate is only 60% when the ionization window is not opened, because the ionization rod is blocked, resulting in ion scattering).

[0139] The reset process after ionization may include: after neutralization, the controller sequentially executes the following steps: the high-voltage generator is de-energized (the ionization rod stops producing ions); the stepper motor drives the ionization window to slide in the reverse direction to the closed position (after confirmation by the position sensor, a low level is output); the ionization section is de-energized, completing the entire process.

[0140] In summary, this embodiment, through the front-opening design of the ionization window, overcomes the limitation of structural obstruction on ion projection, and achieves directional, efficient, and safe image removal in the first-level confidentiality document processing, further improving the physical security defense of high-density printing.

[0141] In some embodiments, the method further includes:

[0142] When the ionization rod is ionized, the temperature near the ionization rod is detected;

[0143] When the temperature rises from a first temperature to a second temperature and then falls to a third temperature, the ionization of the ionization rod is stopped; the first temperature is the temperature before the ionization of the ionization rod; the second temperature is higher than a preset temperature; and the third temperature is lower than the preset temperature.

[0144] After the ionization rod stops ionizing, the ionization window is closed;

[0145] Power on the printing unit and turn it on.

[0146] In some embodiments, ionization initiation and temperature monitoring initialization may include: after the first security level document is printed, the printing section is powered off, and the ionization window is opened, the controller initiates the ionization process: recording the first temperature (T1) before ionization: collected by a temperature sensor, at which time the ionization rod is not charged, and the temperature is consistent with the internal environment of the printer (e.g., 25°C); the high voltage generator is started, and the ionization rod begins to ionize under high voltage (generating positive and negative ions to neutralize the toner cartridge image residue), and multiple temperature sensors simultaneously collect temperature data.

[0147] In some embodiments, temperature curve monitoring and ionization stop determination may include: during ionization, temperature changes follow a heating-peak-cooling curve. Heating phase: After 3 seconds of ionization, the temperature rises from 25°C (T1) to 70°C (T2). At this point, T2 (70°C) is higher than the preset safe temperature (60°C), and the controller continues to monitor (without immediately stopping, as sufficient ion quantity is required for afterimage neutralization). Cooling phase: After 5 seconds of ionization, as the ion generation efficiency stabilizes, the temperature drops from 70°C to 50°C (T3). At this point, T3 (50°C) is lower than the preset temperature (60°C), and the temperature curve shows a continuous downward trend, indicating that the afterimage has been sufficiently neutralized. The controller sends a command to stop ionization: the high-voltage generator is de-energized, and the ionization rod stops producing ions (total ionization time 5 seconds).

[0148] In some embodiments, closing the ionization window and powering on the printing section may include: after ionization stops, a stepper motor drives the ionization window to slide from the open position to the closed position (taking 4 seconds), and the position sensor confirms the closure and sends a feedback signal. After receiving the closure signal, the controller triggers the power supply module of the printing section: components such as the laser and toner cartridge are powered on, and within 3 seconds, it returns to standby mode (toner cartridge voltage -200V, laser ready), waiting for the next printing task.

[0149] In some embodiments, safety assurance during the ionization process may include: monitoring the temperature curve (T1→T2→T3) to prevent tip oxidation caused by continuous operation of the ionization rod at high temperatures (T2>60℃) (increasing lifespan by 20%), while ensuring that the process stops after the temperature drops to a safe range (T3<60℃), balancing image retention (neutralization rate ≥99.9%) and device durability. The connection time between closing the ionization window and powering on the printing section is ≤7s (4s window closing + 3s power on), which reduces device readiness time by 50% compared to a delayed power-on scheme after ionization, adapting to high-frequency, high-density printing scenarios (such as continuous printing needs of classified departments). Temperature monitoring data can be synchronized to the printer's visual feedback system (such as an image of the drum surface captured by a second sensor), ensuring ionization effectiveness through dual verification of the temperature curve and image retention (false positive rate reduced to below 0.5%), and improving the overall security control of high-density documents.

[0150] In summary, this embodiment achieves precise control of the ionization process through dynamic temperature monitoring. Combined with the closing of the ionization window and rapid reset of the printing section, it improves the continuity and efficiency of high-density printing scenarios while ensuring equipment safety and image retention removal.

[0151] In some embodiments, the method further includes adaptively adjusting the printing parameters of the laser printer based on user preferences, the status of the laser printer, its capabilities, or one or more other factors. For example, when the user prefers an ink-saving mode, printing specific sub-pixel features (such as 10-15μm dot matrix) is achieved by inversely enhancing exposure to save on laser printing consumables (such as toner consumption). For pixel-level features or features at or above the pixel level, reducing exposure intermittently and inversely enhancing exposure when the ink-saving mode is enabled is necessary to reduce consumable consumption. This is due to the critical effect of printing sub-pixel-level features. Printing sub-pixel-level features can be used for high-requirement document printing such as anti-counterfeiting tickets and medical images.

[0152] In some embodiments, when the user's preferred quality level is higher than a specified level, the calculation weight of subpixel-level features is increased when determining printing parameters to calculate more suitable printing parameters.

[0153] In one embodiment, an AI model (such as CNN / Transformer) is used to analyze the subpixel-level characteristics of a file (such as gradient distribution and edge energy), not limited to text / image ternary classification. For example, recognizing gradient images requires satisfying the Laplacian smoothness constraint. Dynamically increase the dot matrix density.

[0154] Co-mapping of preferences and features corresponds to the coordinates of the image. The laser energy at the corresponding edge can be as follows:

[0155] Establish a joint mapping function between user preferences (such as high contrast) and pixel features (such as handwritten stroke energy). ,in, For user preference matrix, Pixel edge energy.

[0156] In another embodiment, a laser printer parameter adaptive adjustment method includes:

[0157] Obtaining the user preference matrix ;

[0158] Identify the feature matrix corresponding to the first feature using an AI model. ;

[0159] Real-time monitoring of the laser printer's status, for example, real-time monitoring of the laser printer's hardware status matrix. . It can be used to represent laser power deviation, photosensitive drum charge attenuation rate, fixing temperature fluctuation rate, scanning position error, etc.

[0160] Fusion based on dynamic weighting coefficients α, β, γ , , Generate optimized printing parameters.

[0161] In some embodiments, Table 2 includes a matrix characterizing the current state of the laser printer.

[0162] Table 2

[0163]

[0164]

[0165] Calculated dynamically from the attenuation sensitivity model. The value of i can be 1, 2, 3, or 4. For example, (e.g., when the photosensitive drum wears down, the attenuation sensitivity is increased). Weights).

[0166] For example, fusion based on dynamic weighting coefficients α, β, γ , , The optimized printing parameters can be generated using the following functional relationship:

[0167] .in, These are the optimized printing parameters. : Hardware state matrix (real-time sensor data). α, β, γ: Dynamic weight coefficients, which can be dynamically adjusted through reinforcement learning, or they can be static parameters. : Parameter optimization function, for example, This can represent a neural network mapping layer. In some embodiments, the dynamic weight coefficients are updated through a reinforcement learning model, and the reward function is: ,in, For print quality, For printing energy consumption, This refers to the status changes of the laser printer.

[0168] Parameter adjustments between pages may include at least one of the following when any of the following conditions are met:

[0169] Content type switching: The content characteristics of two consecutive pages change significantly (e.g., page 1 is plain text, page 2 is an RGB image).

[0170] User preference changes: Users can adjust their preference settings in real time during the printing process via the APP (such as switching from standard mode to high quality mode).

[0171] Hardware status warning: The sensor detects a sudden change in the status of a critical component (such as a sudden increase in the temperature of the photosensitive drum by 5°C or a fluctuation in laser power exceeding ±5%).

[0172] Substrate type change: The paper thickness / type detector detects a change in the media properties of the next page (e.g., switching from 80g plain paper to 200g coated paper).

[0173] Preload analysis: When printing page N, pre-parse the PCL / PS data stream of page N+1 and identify content characteristics (text / image / mixed type) through an AI model.

[0174] Parameter switching strategy: If the difference in characteristics between adjacent pages exceeds a threshold (e.g., SSIM (structural similarity) < 0.6), a print parameter reset is triggered. Specifically, new print parameters are loaded during the page gap. For user preference changes: The current transmission queue is immediately interrupted, and a parameter update command is inserted. Hardware coordination ensures: Laser power calibration and fixing temperature adjustment are performed during the page gap to avoid switching parameters within the print line.

[0175] Switching delay control: Cross-page parameter switching must be completed within the page turning interval (e.g., <50ms) to avoid affecting throughput.

[0176] In some embodiments, parameter adjustments can be made across regions within a single page, specifically when the following conditions are met simultaneously within a single page: significant differences in regional characteristics: text regions and image regions are adjacent (e.g., title + photo combination), and the resolution difference is greater than a specified value, such as 200 dpi.

[0177] There is a coexistence of gradient regions and RGB image regions (requiring different halftone algorithms).

[0178] Limited local scanning capabilities of laser printers: The scanning motor cannot maintain high-precision positioning when moving at high speeds (the local scanning speed needs to be reduced).

[0179] Region segmentation and parameter mapping: An AI model is used to divide a printed page into several regions (such as a title area, image area, and background area), and feature labels are generated for each region.

[0180] Dynamic laser control: Based on the feature label, the printing parameters are adaptively switched by region during the scanning process.

[0181] Adaptive halftone switching algorithm:

[0182] Text area: Use raster binarization to preserve sharp edges;

[0183] Gradient region: Switch to error diffusion algorithm to prevent color banding.

[0184] Dynamic scanning speed reduction: Automatically reduces the scanning speed in handwritten text areas with high curvature. Formula:

[0185] ;in, This is the adjusted scan rate. The basic scan rate for printing the current page. The curvature of the current text stroke. The maximum curvature of the stroke.

[0186] In some embodiments, hardware provides real-time feedback: an integrated photoelectric sensor detects the actual imaging effect at the end of the scan line. If ghosting / blurring is detected at the boundary of the region, compensation is immediately triggered: the number of scan repetitions at the boundary is increased, and the laser power at the boundary is finely adjusted.

[0187] In some embodiments, when the parameter differences between adjacent areas within the same page are too large (e.g., laser power difference > 30%), a transition buffer (2-3 scan lines) is inserted, and linear interpolation is used to smooth the transition parameters.

[0188] like Figure 3 As shown, this disclosure provides an information security control device for using a laser printer. The laser printer includes a printing section and an ionization section. The printing section includes a toner cartridge. The ionization section includes ionization rods arranged parallel to the toner cartridge. The printing section and the ionization section are electrically isolated. The device includes:

[0189] The determination module 3110 is used to identify the text content of the first document using the recognition layer of the locally deployed AI model; and to determine the security level of the first document to be printed based on the identified text content using the prediction layer of the locally deployed AI model.

[0190] The writing module 3120 is used to write the first data of the first file to the print buffer of the laser printer and prevent the first data of the first file from being written to the hard disk of the laser printer when the security level of the first file is the first security level.

[0191] Printing module 3130 is used to print the first file based on the first data stored in the print cache;

[0192] The overlay module 3140 is used to write preset data into the print cache after the first file is printed to overwrite the first data in the print cache;

[0193] The power-off module 3150 is used to power off the printing section after the print buffer has been overwritten;

[0194] The neutralization module 3160 is used to charge the ionization rod after the printing section is powered off, so that the ionization rod generates ions to neutralize the residual image on the drum; an isolation layer is provided on the side of the ionization rod away from the drum; the isolation layer is used to protect the circuitry inside the laser printer.

[0195] In some embodiments, the writing module is further configured to write the first data of the first file to the print cache of the laser printer and prevent the first data of the first file from being written to the hard disk of the laser printer when the security level of the first file is the second security level; the second security level is lower than the first security level.

[0196] After the first file is printed, the overlay module writes preset data into the print cache to overwrite the first data in the print cache.

[0197] The device further includes:

[0198] A maintenance module is used to keep the printing section powered on.

[0199] In some embodiments, the writing module is further configured to write the first data of the first file to the print cache of the laser printer and allow the first data of the first file to be written to the hard disk of the laser printer when the security level of the first file is the third security level; the third security level is lower than the second security level.

[0200] The overlay module is further configured to overwrite the first data stored in the print cache with the second data of the second file to be printed.

[0201] Combination Figure 4 As shown, this application embodiment provides an electronic device including a processor 10 and a memory 11. Optionally, the device may further include a communication interface 12 and a bus 9. The processor 10, communication interface 12, and memory 11 can communicate with each other via the bus 9. The communication interface 12 can be used for information transmission. The processor 10 can call logical instructions in the memory 11 to execute the information security control method for using a laser printer as described in the above embodiment.

[0202] Furthermore, the logical instructions in the aforementioned memory 11 can be implemented as software functional units and, when sold or used as independent products, can be stored in a computer-readable storage medium.

[0203] The memory 11, as a computer-readable storage medium, can be used to store software programs and computer-executable programs, such as program instructions / modules corresponding to the methods in the embodiments of this application. The processor 10 executes functional applications and data processing by running the program instructions / modules stored in the memory 11, thereby implementing the information security control method for using a laser printer in the above embodiments.

[0204] The memory 11 may include a program storage area and a data storage area. The program storage area may store the operating system and application programs required for at least one function; the data storage area may store data created based on the use of the electronic device. Furthermore, the memory 11 may include high-speed random access memory and may also include non-volatile memory.

[0205] This application provides a computer program product, which includes a computer program stored on a storage medium. The computer program includes program instructions, which, when executed by a computer, cause the computer to perform the aforementioned information security control method for using a laser printer.

[0206] The aforementioned computer-readable storage medium may be a transient computer-readable storage medium or a non-transitory computer-readable storage medium.

[0207] The technical solutions of this application embodiment can be embodied in the form of a software product. This computer software product is stored in a storage medium and includes one or more instructions to cause a computer device (which may be a personal computer, server, or network device, etc.) to execute all or part of the steps of the method described in this application embodiment. The aforementioned storage medium can be a non-transitory storage medium, including various media capable of storing program code such as USB flash drives, portable hard drives, read-only memory, random access memory, magnetic disks, or optical disks, or it can be a transient storage medium.

[0208] In the above embodiments of this application, the descriptions of each embodiment have different focuses. For parts not described in detail in a certain embodiment, please refer to the relevant descriptions of other embodiments.

[0209] The embodiments or examples disclosed in this application are not exhaustive, but merely illustrative of some embodiments or examples, and are not intended to limit the scope of protection of this disclosure. Unless contradictory, each step in a particular embodiment or example can be implemented as an independent embodiment, and the steps can be arbitrarily combined. For example, a solution after removing some steps in a particular embodiment or example can also be implemented as an independent embodiment, and the order of the steps in a particular embodiment or example can be arbitrarily interchanged. Furthermore, optional methods or examples in a particular embodiment or example can be arbitrarily combined; moreover, embodiments or examples can be arbitrarily combined. For example, some or all steps of different embodiments or examples can be arbitrarily combined, and a particular embodiment or example can be arbitrarily combined with optional methods or examples of other embodiments or examples.

[0210] In the several embodiments provided in this application, it should be understood that the disclosed technical content can be implemented in other ways. The device embodiments described above are merely illustrative; for example, the division of units can be a logical functional division, and in actual implementation, there may be other division methods. For instance, multiple units or components may be combined or integrated into another system, or some features may be ignored or not executed. Furthermore, the displayed or discussed mutual coupling, direct coupling, or communication connection may be through some interfaces; the indirect coupling or communication connection between units or modules may be electrical or other forms.

[0211] The units described as separate components may or may not be physically separate. The components shown as units may or may not be physical units; that is, they may be located in one place or distributed across multiple units. Some or all of the units can be selected to achieve the purpose of this embodiment according to actual needs.

[0212] Furthermore, the functional units in the various embodiments of this application can be integrated into one processing unit, or each unit can exist physically separately, or two or more units can be integrated into one unit. The integrated unit can be implemented in hardware or as a software functional unit.

[0213] If the integrated unit is implemented as a software functional unit and sold or used as an independent product, it can be stored in a computer-readable storage medium. Based on this understanding, the technical solution of this application, in essence, or the part that contributes to the prior art, or all or part of the technical solution, can be embodied in the form of a software product. This computer software product is stored in a storage medium and includes several instructions to cause a computer device (which may be a personal computer, server, or network device, etc.) to execute all or part of the steps of the methods described in the various embodiments of this application. The aforementioned storage medium includes various media capable of storing program code, such as a USB flash drive, read-only memory (ROM), random access memory (RAM), portable hard drive, magnetic disk, or optical disk.

[0214] The above description is only a preferred embodiment of this application. It should be noted that for those skilled in the art, several improvements and modifications can be made without departing from the principle of this application, and these improvements and modifications should also be considered within the scope of protection of this application.

Claims

1. An information security control method for using a laser printer, characterized in that, The laser printer includes a printing section and an ionization section; the printing section includes a toner cartridge; the ionization section includes ionization rods arranged parallel to the toner cartridge. The printed portion and the ionized portion are electrically isolated; the method includes: The recognition layer of the locally deployed artificial intelligence (AI) model is used to identify the text content of the first document; The prediction layer of the locally deployed AI model determines the security level of the first document to be printed based on the recognized text content; When the security level of the first file is the first security level, the first data of the first file is written to the print cache of the laser printer and the writing of the first data of the first file to the hard disk of the laser printer is prohibited. Based on the first data stored in the print cache, print the first file; After the first file is printed, preset data is written to the print cache to overwrite the first data in the print cache; After the print buffer has been overwritten, power is turned off to the printing section. After the printing section is powered off, the ionization rod is charged, causing the ionization rod to generate ions that neutralize the residual image on the drum; an isolation layer is provided on the side of the ionization rod away from the drum; the isolation layer is used to protect the circuitry inside the laser printer.

2. The method according to claim 1, characterized in that, The recognition layer of the AI ​​model includes a first recognition unit, a prediction unit, and a second recognition unit connected in series; the recognition layer of the locally deployed AI model identifies the text content of the first document, including: The first recognition unit processes the text content with a first scale parameter to obtain a first recognition result; The prediction unit predicts the recognition direction based on the first recognition result; The second recognition unit adjusts the first scale parameter according to the recognition direction to obtain the second scale parameter; The first recognition unit processes the text content based on the second scale parameter to obtain a second recognition result, and fuses the first recognition result and the second recognition result to obtain a recognition result that matches the security level.

3. The method according to claim 1 or 2, characterized in that, The method further includes: When the security level of the first file is the second security level, the first data of the first file is written to the print buffer of the laser printer and the writing of the first data of the first file to the hard disk of the laser printer is prohibited; the second security level is lower than the first security level. After the first file is printed, preset data is written to the print cache to overwrite the first data in the print cache; Keep the printing unit powered on.

4. The method according to claim 3, characterized in that, The method further includes: When the security level of the first file is the third security level, the first data of the first file is written to the print buffer of the laser printer and the first data of the first file is allowed to be written to the hard disk of the laser printer; the third security level is lower than the second security level. The first data stored in the print cache is overwritten with the second data from the second file to be printed.

5. The method according to claim 1 or 2, characterized in that, The method further includes: When the security level is the third security level, the first data of the first file is written to the print cache and / or hard disk according to the current state of the laser printer.

6. The method according to claim 5, characterized in that, The method further includes: Before charging the ionization rod, the ionization window is opened to expose the ionization rod, which is arranged in parallel with the drum.

7. The method according to claim 6, characterized in that, The method further includes: When the ionization rod is ionized, the temperature near the ionization rod is detected; When the temperature rises from a first temperature to a second temperature and then falls to a third temperature, the ionization of the ionization rod is stopped; the first temperature is the temperature before the ionization of the ionization rod; the second temperature is higher than a preset temperature; and the third temperature is lower than the preset temperature. After the ionization rod stops ionizing, the ionization window is closed; Power on the printing unit and turn it on.

8. An information security control device for using a laser printer, characterized in that, The laser printer includes a printing section and an ionization section; the printing section includes a toner cartridge; the ionization section includes an ionization rod arranged parallel to the toner cartridge; the printing section and the ionization section are electrically isolated; the device includes: The determination module is used to identify the text content of the first document using the recognition layer of the locally deployed AI model; and to determine the security level of the first document to be printed based on the identified text content using the prediction layer of the locally deployed AI model. The writing module is used to write the first data of the first file to the print buffer of the laser printer and prevent the first data of the first file from being written to the hard disk of the laser printer when the security level of the first file is the first security level. The printing module is used to print the first file based on the first data stored in the print cache; An overwrite module is used to write preset data into the print cache after the first file is printed to overwrite the first data in the print cache; A power-off module is used to cut off power to the printing section after the print buffer has been overwritten; A neutralization module is used to charge the ionization rod after the printing section is powered off, so that the ionization rod generates ions to neutralize the residual image on the drum; an isolation layer is provided on the side of the ionization rod away from the drum; the isolation layer is used to protect the circuitry inside the laser printer.

9. The apparatus according to claim 8, characterized in that, The writing module is further configured to write the first data of the first file to the print cache of the laser printer and prevent the first data of the first file from being written to the hard disk of the laser printer when the security level of the first file is the second security level; The second security level is lower than the first security level; After the first file is printed, the overlay module writes preset data into the print cache to overwrite the first data in the print cache. The device further includes: A maintenance module is used to keep the printing section powered on.

10. The apparatus according to claim 9, characterized in that, The writing module is further configured to write the first data of the first file to the print buffer of the laser printer and allow the first data of the first file to be written to the hard disk of the laser printer when the security level of the first file is the third security level; the third security level is lower than the second security level. The overlay module is further configured to overwrite the first data stored in the print cache with the second data of the second file to be printed.

Images