Method for processing after-reading-burning multimedia content based on fuzzy preview and long-press unlocking

Abstract

The application provides a kind of based on fuzzy preview and long press unlocking after reading multimedia content processing method, belongs to instant messaging and data security field technical field;In the sending end, the target content is uniformly identified, and is transmitted to the receiving end through an end-to-end encryption channel.The receiving end enters the safe preview mode after analyzing the message, and the content is blurred or covered;The user needs to long press in the specified transition time to gradually remove the protection, and can enter the clear state.In the clear state, the system starts the destruction timer, supports timed and immediate destruction;At the same time, it contains local peek mode and panoramic safety mode, and suppresses shooting and transmission.In the receiving end, the return receipt with digital signature is returned after the destruction is completed.The application can maintain good user experience while significantly improving the security and reliability of the after-reading-burning function for pictures, text and videos, and has wide application prospect.

Description

Technical Field

[0001] This invention belongs to the technical field of instant messaging and data security, and in particular relates to a method for processing self-destructing multimedia content based on fuzzy preview and long press unlock. Background Technology

[0002] With the rapid development of mobile internet, instant messaging (IM) software has become a primary means of daily communication. Users exchange not only text messages through IM tools, but also a large amount of multimedia data such as images and videos. Especially when dealing with sensitive information such as contract signings, financial documents, medical reports, and work results, users' demand for messages that are "temporary, one-time, and unreproducible" is becoming increasingly strong.

[0003] To meet these needs, some instant messaging platforms have introduced a "self-destructing message" feature. The basic idea is that once a message is read by the recipient, it is destroyed shortly afterward and is no longer stored on the device or server. However, existing self-destructing message technology still has the following shortcomings in multimedia scenarios:

[0004] (1) The timed deletion is simple and lacks multi-layer protection.

[0005] Most solutions employ a single strategy of "timeout upon opening → deletion upon expiration." During the timeout phase, the content is typically in a fully clear state; images can be screenshotted / taken photos, text can be read or copied directly, and videos can be screen-captured or recorded, thus weakening the actual constraint of "one-time display."

[0006] (2) Lack of anti-accidental touch design, resulting in inconsistency between user experience and security boundaries.

[0007] Many apps allow users to access a clear interface simply by clicking on a thumbnail, without setting a clear "secondary confirmation / continuation intent" action. This can lead to accidental triggering of the post-view process, wasting the only viewing opportunity and easily causing communication misunderstandings and security disputes.

[0008] (3) It lacks the ability to prevent screenshots and screen recording, and it is difficult to cover external devices for shooting.

[0009] System-level restrictions on screenshots / screen recordings vary significantly across different operating systems and versions, and can be bypassed; while external camera captures (screen recordings) are almost impossible to prevent through "screenshot ban" measures.

[0010] (4) Weak control over text and system cache residues.

[0011] Text content can often be copied, saved, or remain in system components such as rendering cache, clipboard, fonts / bitmap textures; thumbnails of images and videos, recent media indexes, and decoding buffers may also be recoverable outside the application, leading to the risk of "superficial deletion, underlying residue".

[0012] (5) Receipt loss and status asynchrony in weak network environment.

[0013] When the network is unstable or temporarily interrupted, the read receipt may not be delivered in a timely manner, and the sender may not be able to confirm whether the receiver has completed the destruction, resulting in inconsistencies in status and difficulties in obtaining evidence.

[0014] (6) Insufficient cross-platform consistency.

[0015] Different operating systems vary significantly in rendering / decoding / storage paths, permissions, and API capabilities. Existing solutions often only adapt to a single platform, lacking a unified interface abstraction and performance / security baseline, making it difficult to scale up and maintain in the long term.

[0016] (7) There is a lack of proactive measures to suppress the circulation of photographs and to trace their origins.

[0017] Most solutions rely solely on "screenshot bans" and do not overlay traceable digital watermarks during the clear display stage, making it difficult to locate the source and determine responsibility for leaked files after taking photos / screen captures.

[0018] (8) The terminal-side buffer / video memory / temporary file cleanup is incomplete.

[0019] Media data may exist in multiple copies in memory, video memory textures, decoding queues, temporary files, and logs; if only "logical deletion" is performed without encryption key cleanup and minimum resident control, there is still a risk that it can be recovered by tools.

[0020] (9) Insufficient flexibility and configurability.

[0021] Key parameters such as viewing duration, unlocking method, destruction strategy, and retransmission receipt are mostly fixed values ​​or weakly configurable, making it difficult to make fine-grained adjustments according to business scenarios (such as office / medical / social), thus limiting actual usability.

[0022] In recent years, academia and industry have attempted to improve security by introducing biometrics, key management, and trusted execution environments. However, these approaches have encountered challenges such as complex implementation, high resource consumption, and difficulties in cross-platform migration. Overall, mainstream applications still exhibit fragmented functionalities, rigid interactions, and a lack of integrated design encompassing user experience, technical implementation, and system fault tolerance. Summary of the Invention

[0023] To address the aforementioned problems, this invention proposes a method for processing self-destructing multimedia content based on fuzzy preview and long-press unlock, comprising the following steps:

[0024] When a user selects target content, the sending end marks the content as self-destructing after reading. The mark includes at least filename modification and protocol field addition.

[0025] The receiving end performs a security preview process during the initial display: Gaussian blur is applied to images, graphical rendering and masking are applied to text, and preview frames are blurred or their sharpness is reduced for videos.

[0026] During the security preview phase, two security modes are provided: a partial peeping mode and a full-view security mode. In partial peeping mode, when a user touches the screen with a single finger, the coordinates of the touch point are obtained in real time, and a clear view is rendered centered on that point, while the surrounding area remains blurred, allowing the user to partially view the content. In full-view security mode, the system detects whether there are two touch points with a distance between them that is not less than a preset threshold. If the continuous pressing time reaches the preset transition duration, the system switches to full-view clear display. During the transition, if any finger is released, the touch point moves off the screen, the application is switched to the background, or the touch point jitter exceeds the tolerance range, the system immediately reverts to the security preview state without consuming viewing opportunities.

[0027] Once the system enters clear display / playback mode, it starts a timer to display / play content according to the configured duration, and immediately destroys the content when the user releases their finger or the timer ends.

[0028] After the destruction is complete, the receiving end generates a digitally signed receipt message and returns it to the sending end.

[0029] Preferably, after parsing the self-destructing message, the receiving end establishes a local access status record, which includes at least a unique content identifier, viewing status, receiving timestamp, and storage path; the multimedia content is encrypted and stored locally using a symmetric encryption key dynamically generated by the session, the key itself is protected through the platform security interface, and is bound to the device fingerprint, content hash, and policy parameters.

[0030] Preferably, the Gaussian blur applied to the image has a blur radius that is dynamically adjusted non-linearly based on the screen pixel density (DPI) and content features, specifically including:

[0031] The basic blur radius is determined based on the screen DPI of the receiving device. And based on the content complexity factor of the target image and preset security level The final blur radius is calculated using the following formula:

[0032] ;

[0033] in and These are the weighting coefficients. , , The content complexity factor Quantization is achieved by calculating the contrast and energy values ​​of the image's gray-level co-occurrence matrix (GLCM). The value ranges from 0.5 to 1.5; the simpler the image content, the better. The larger the value, the higher the degree of ambiguity; the preset security level Defined by the sender in the message protocol field, the value ranges from 1 to 3. The higher the value, the higher the ambiguity.

[0034] Preferably, the Gaussian blur applied to the image, in safe preview mode, achieves the blur effect through a two-dimensional Gaussian kernel function, the standard deviation of which is... With the final blur radius The relationship is ,in is a proportionality constant, ranging from 2.5 to 3.5; the formula for the two-dimensional Gaussian kernel function is:

[0035] ;

[0036] in, Indicates the convolution kernel in coordinates The weight value at the location; and These are the horizontal and vertical offsets relative to the center point of the convolution, in pixels. It is a natural exponential function.

[0037] Preferably, the basic fuzzy radius The DPI of the receiving device is determined piecewise linearly: when DPI ≥ 480, ;When 320 ≤ DPI hour, ;when DPI hour, Where 200 is the preset minimum DPI threshold, when DPI hour, The blur radius is fixed at 20px; the blur transition time is 0.5–1 second; the blur radius gradually decreases to 0 by linear or exponential interpolation; and the user's interface displays the message "Release to destroy" during the unlocking process.

[0038] Preferably, in panoramic security mode, the blur removal process follows a non-linear transition curve, with a blur radius... The change with time t is defined by the following formula:

[0039] ;

[0040] in, The duration of continuous pressure applied by the user. ; The preset total transition time ranges from 0.5 to 1.5 seconds. This is the transition coefficient, ranging from 0.8 to 2.0; when At that time, the fuzzy resolution process is fast at first and then slows down; when At that time, the fuzzy resolution process is slow at first and then fast; when When, it is a linear release; when hour, Enter clear display mode.

[0041] Preferably, a clear phase is overlaid with a dual-layer digital watermark (bright / dark) to suppress the transmission of images and enable source tracing. The bright watermark is a semi-transparent text or graphic containing a device fingerprint, timestamp, and random sequence. The dark watermark is embedded through spatial domain LSB or frequency domain DCT / DFT perturbation and bound to the device fingerprint and random number to enable source tracing of the captured or forwarded images.

[0042] Preferably, the destruction mechanism includes the secure destruction of the encryption key, specifically: deleting the key entry by calling the Keychain interface on the iOS platform, destroying the key by calling Keystore / KeyMint on the Android platform, and destroying the key by calling the Cryptographic API or TPM-related interfaces on the Windows / Linux platform; after the key is destroyed, even if the locally encrypted content file still exists, it cannot be decrypted and recovered.

[0043] Preferably, in weak network or network outage scenarios, the signature receipt message generated by the receiving end is written to the local queue cache, and the queue adopts a FIFO strategy; the client checks the network status every 30 seconds, and automatically retransmits the queue message after recovery; the retransmission mechanism adopts an exponential backoff strategy, with retry intervals of 30 seconds, 60 seconds, 120 seconds up to 10 minutes, and if the transmission fails for more than 72 hours, it is discarded and an exception log is recorded.

[0044] Preferably, the client retrieves the latest configuration parameters from the cloud upon startup or when switching to the foreground. These parameters include the fuzzy radius range, viewing duration, unlocking method, destruction strategy, watermark strength, and retransmission strategy. The client performs digital signature verification on the retrieved configuration parameters, and stores them in a local read-only cache after successful verification. When the network is unavailable, the client uses the previously valid configuration parameters for subsequent processing.

[0045] The innovative aspects and beneficial effects of this invention include:

[0046] By using a two-level interaction mechanism of "fuzzy preview + long press to unlock", the system effectively avoids users accidentally consuming viewing opportunities and significantly reduces the risk of being watched and leaked in public places, thus solving the problems of unreasonable interaction design and unclear security boundaries in existing solutions.

[0047] By employing a strategy that includes cleaning up decrypted data in memory and video memory and securely destroying encryption keys, combined with minimum resident buffer control, the system ensures that content data is completely and irreversibly deleted, overcoming the risk of data residue caused by logical deletion in existing technologies.

[0048] By overlaying a dual-layer digital watermark (bright / dark) bound to the device and session during the clear display phase, along with partial peeping mode and panoramic security mode, the system effectively suppresses the shooting and secondary dissemination of screen content without affecting the user's viewing experience. It also provides a technical basis for tracing the source of the leak and makes up for the shortcomings of existing solutions in preventing shooting.

[0049] By using digital signature receipts and an automatic retransmission mechanism based on exponential backoff, the sender can reliably verify the receiver's destruction behavior and ensure the eventual consistency of the states at both ends even when the network is unstable. This solves the problems of receipt loss and state asynchrony in existing solutions under weak network conditions. Attached Figure Description

[0050] Figure 1 This is a flowchart illustrating the overall implementation logic of the present invention;

[0051] Figure 2 This is an interactive diagram illustrating how the present invention allows users to gradually enter a clear display mode by long-pressing a multimedia thumbnail.

[0052] Figure 3 This is a schematic diagram of the countdown prompt interface clearly displayed during the period of the present invention;

[0053] Figure 4 This is a schematic diagram of the interface where a placeholder block is displayed in the chat window after the multimedia file is destroyed according to the present invention.

[0054] Figure 5 This is an overall flowchart of the interaction between the self-destructing multimedia processing method of the present invention and the message pipeline and feedback channel;

[0055] Figure 6 This is a flowchart illustrating the logic of local caching and automatic retransmission of receipt messages in weak network scenarios according to the present invention.

[0056] Figure 7 This is a system architecture diagram showing the division of client modules under the cross-platform framework support of this invention. Detailed Implementation

[0057] To make the objectives, technical solutions, and advantages of the embodiments of this application clearer, the technical solutions of the embodiments of this application will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some, not all, of the embodiments of this application. All other embodiments obtained by those skilled in the art based on the embodiments of this application without creative effort are within the scope of protection of this application.

[0058] Example 1:

[0059] This invention provides a method for processing self-destructing multimedia content based on fuzzy preview and long-press unlock. The overall process is as follows: Figure 1 As shown, the interaction details between the message pipeline and the feedback channel are as follows: Figure 5 As shown.

[0060] S1, when the user selects target content, the sending end marks the content as "read-and-burn," and the mark includes at least filename modification and addition of a protocol field; the protocol field carries the security level. View duration Strategy parameters, etc.

[0061] S2, the receiving end performs a security preview process during the initial display: applies a Gaussian blur to the image, with a blur radius of... According to the formula Dynamic calculation, in which Determined linearly based on screen DPI segmentation. The complexity of image content is quantified using a grayscale co-occurrence matrix; text is rendered graphically and masked; and video preview frames are blurred or have reduced clarity.

[0062] Users need to long-press during a specific transition time. The blur / mask is gradually removed internally, and the blur removal process follows a non-linear transition curve. Only then can you enter the clear display / playback mode.

[0063] S3, after entering the clear display / playback mode, the system starts a timer to display / play content according to the configured duration, and immediately destroys it when the user releases their finger or the timer ends. The destruction includes clearing the memory cache, video memory texture, files and video buffer, and securely deleting the encryption key used to decrypt the content; a light / dark dual-layer digital watermark is superimposed in the clear stage.

[0064] S4. After the destruction is completed, the receiving end generates a receipt message with a digital signature and returns it to the sending end. The receipt includes a content hash, a destruction timestamp, and device fingerprint information. In a weak network environment, the receipt is guaranteed to be delivered through a local queue cache and an automatic retransmission mechanism.

[0065] (I) Content Identification and Transmission

[0066] Users select target content (including images, text, and videos) and check the "self-destructing message" option at the sending end. The system assigns a unique, consistent identifier to this content for accurate identification and policy enforcement at both the transmitting and receiving ends. The identifier mechanism covers two levels:

[0067] File / Resource Layer Identifier: Modify the filename or resource identifier (URI) level, such as adding the prefix "flash_" or the suffix ".burn", to ensure that the content can be recognized at the file system level;

[0068] Message protocol layer identifier: Add a flag field (e.g., 1 byte, value set to 0xF1) and a self-describing parameter structure (e.g., TLV format) to the message protocol header. Parameters include: policy version (ver), media type (media_type), viewing duration (ttl, range 3-120 seconds), unlocking method (policy), destruction policy (mode), digital watermark on / off switch and strength (wm_on / wm_strength), minimum video buffer dwell threshold (min_buffer_ms), codec profile (codec_profile), and content hash (content_hash) calculated on the final transmitted bitstream.

[0069] The aforementioned identifier is written into the logical protocol header before end-to-end encryption (E2EE) and transmitted encrypted along with the payload content to ensure security and compatibility. The content hash is calculated based on the bitstream encoded at the sending end and bound to the encoding bit, ensuring that the verification criteria for "the same content" are consistent between the sending and receiving ends, laying the foundation for subsequent reliable receipts and traceability.

[0070] (II) Receiver Identification and Status Establishment

[0071] Upon arrival at the receiving end, the message is decrypted via E2EE and the logical protocol header is parsed. When the flag field 0xF1 is detected, it is confirmed as a self-destructing message, and all policy parameters are read. The receiving end immediately creates an access status record in a local lightweight database (such as SQLite) using the content hash as a unique identifier (uid). Record fields include: uid, media type, viewing status (initially "not viewed"), reception timestamp (recv_time), policy parameters, and local sandbox storage path. The text content is first rendered as a bitmap and its hash is calculated. All status and cache files are strictly stored in the application's private directory, with no backup and no indexing flags set to achieve system-level isolation. This step establishes verifiable access control boundaries.

[0072] When saving multimedia content, the receiving end uses a session-based dynamic key to encrypt and store the original data. Specifically, the system generates a unique symmetric encryption key (such as AES-256-GCM), uses this key to encrypt image data, and stores it in a local private directory. The encryption key itself is protected and stored through the platform's secure key management interface and is bound to the device fingerprint, content hash, and policy parameters of the current session. This mechanism ensures that even if the locally encrypted file is extracted, the original content cannot be decrypted without the key.

[0073] (III) Security Preview and Long Press to Unlock

[0074] When a user clicks on this content for the first time in the chat interface, the clear content is not displayed directly, but instead enters a secure preview mode:

[0075] Image: Gaussian blur processing was performed. The blur radius was calculated using an adaptive dynamic algorithm, specifically including the following steps:

[0076] Step 1: Determine the basic fuzzy radius

[0077] The system obtains the screen DPI of the receiving device and determines the basic ambiguity radius based on a piecewise linear function. :

[0078] When DPI ≥ 480, ;

[0079] When 320 ≤ DPI hour, ;

[0080] When DPI hour, Where 200 is the preset minimum DPI threshold, when DPI hour, Fixed at 20px.

[0081] Step 2: Calculate the content complexity factor

[0082] The system performs real-time analysis of image content and calculates the content complexity factor. The specific method is as follows: convert the image to grayscale, calculate the gray-level co-occurrence matrix (GLCM), and extract two feature quantities: contrast and energy.

[0083] ;

[0084] in The gray values ​​in the gray-level co-occurrence matrix and The probability of co-occurrence. Normalizing and weighting contrast and energy, we obtain... The value ranges from 0.5 to 1.5. The simpler the image content, the better. The higher the value, the greater the blurriness; the more complex the image content, The smaller the value, the lower the degree of ambiguity, thus retaining more effective information during the preview stage.

[0085] Step 3: Calculation of final fuzzy radius

[0086] The system reads the preset security level from the message protocol field. (Values ​​range from 1 to 3, with higher values ​​indicating higher security requirements.) Combining the results of steps 1 and 2, calculate the final fuzzy radius using the following formula. :

[0087] ;

[0088] Among them, the weighting coefficient and satisfy , , In this embodiment, we take... , .

[0089] Step 4: Implement Gaussian blur

[0090] The image is convolved using a two-dimensional Gaussian kernel function, and the standard deviation of the Gaussian kernel is... With the final blur radius The relationship is The proportionality constant The value range is 2.5 to 3.5; in this embodiment, it is taken as... The formula for the two-dimensional Gaussian kernel function is:

[0091] ;

[0092] in Indicates the convolution kernel in coordinates The weight value at that point, and These are the horizontal and vertical offsets relative to the center point of the convolution, respectively (in pixels). It is a natural exponential function.

[0093] Text: First render as a texture or bitmap, then overlay a semi-transparent mask and disable the system copy function.

[0094] Video: Only the first frame or a small number of preview frames are extracted and blurred or reduced in resolution.

[0095] The preview interface prompts "Long press to view". Users need to continuously press the preview area, and the blur effect or mask will gradually weaken within the set transition time (T, usually 0.5-1.0 seconds).

[0096] For the image blur removal process, the system uses a non-linear transition curve with a blur radius. Over time The change is defined by the following formula:

[0097] ;

[0098] The duration of continuous pressure applied by the user. ; To preset the total transition time, this embodiment takes... Second; This is a transition coefficient, ranging from 0.8 to 2.0. In this embodiment, it is selected based on the social scenario. This achieves a fast-then-slow unlocking experience, allowing users to quickly get a preliminary look at the content, and then slowing down the changes in the final stage to avoid accidental triggering due to slight vibrations.

[0099] when hour, During the same rendering cycle, the system switches to clear display / playback mode. During this transition period ( Clear video frames will not be pushed to the visible buffer. If the user releases the device prematurely or an interruption event occurs such as a phone call or switching to the background, the system will immediately revert to the safe preview state without consuming viewing opportunities. The interface will simultaneously display a circular progress bar and the message "Release to destroy".

[0100] This mechanism significantly reduces the false touch rate, while also utilizing a configurable transition coefficient. It achieves a dynamic balance between security and user experience in different scenarios. The long-press unlock interaction process is as follows: Figure 2 As shown.

[0101] (iv) Clear display, watermark overlay and multi-layer destruction

[0102] After entering clear mode:

[0103] Duration control and instant destruction: The system automatically starts a countdown (duration is set by the TTL parameter, default 30 seconds). Users can also release the key at any time to trigger immediate destruction. The interface displays the remaining time, such as... Figure 3 As shown, this allows users to clearly see the remaining time;

[0104] Digital watermark overlay: While displaying / playing content clearly, the system automatically overlays a dual-layer digital watermark (bright / dark) onto the content. The bright watermark is semi-transparent text / graphics, containing a device fingerprint (truncated), timestamp, and session random factor; the dark watermark embeds the same information through spatial domain (such as LSB) or frequency domain (such as DCT) perturbations. The watermark is used to suppress screen captures and enable source tracing of leaks;

[0105] Complete Destruction: When the countdown ends or the user releases their device, the system performs multi-level, irreversible data destruction: (i) Video Memory: Releases the currently displayed texture and allocates a new texture to cover it. (ii) Memory: Zeroes the decrypted data in Native memory, such as pixel data and decoding buffers (e.g., memset / SecureZeroMemory). (iii) Encryption Key Destruction: Securely deletes the encryption key used to decrypt the content, making the locally encrypted file inaccessible. (iv) Video Buffer: Clears the decoder output queue. (v) Status Update: Updates the local access status record to "Destroyed" and records the destruction timestamp.

[0106] (V) Signature Receipt Generation and Weak Network Retransmission

[0107] After the destruction action is completed, a gray placeholder block will appear in the original content position in the chat window, such as... Figure 4 As shown. Simultaneously, the receiving end generates a digitally signed receipt message. The core fields of the receipt include: content hash (content_hash), destruction timestamp (destroy_time), destruction status code (status), and a privacy-preserving representation of the device fingerprint (device_id_trunc). The receipt is signed using the receiving end's private key (preferably using the ECDSA-P256 algorithm) and sent to the sending end via the existing E2EE channel. The sending end verifies the signature and hash using the corresponding public key. After confirming the destruction is genuine, it updates the interface status (e.g., displaying "Destroyed by the other party").

[0108] For scenarios with weak or no network connectivity, if a delivery receipt fails to be sent, it will be persisted to a local queue. The client's background service will periodically (e.g., every 30 seconds) attempt to resend the receipt, employing an exponential backoff strategy (gradually increasing retry intervals) to avoid network congestion. A 72-hour limit is typically set; receipts exceeding this timeout are discarded and logged to ensure eventual consistency.

[0109] (vi) Cross-platform compatibility and dynamic configuration

[0110] To achieve consistent user experience and security across different operating systems such as Android, iOS, Windows, Linux, and HarmonyOS, this invention designs:

[0111] Interface Abstraction Layer: Defines a set of platform-independent APIs, such as applyBlur, watermark, securePlay, and secureErase. Each platform only needs to implement the underlying calls to these interfaces according to its own characteristics (e.g., Android uses OpenGL ES, iOS uses Metal, and Windows uses Skia).

[0112] Dynamic cloud configuration: Key parameters (such as blur radius range, viewing duration, unlocking method, destruction strategy, watermark strength, re-upload strategy, etc.) can be sent to the cloud in JSON format. After the client pulls the configuration, it can dynamically adjust its behavior without updating the application, greatly improving the flexibility and maintainability of the solution.

[0113] This embodiment demonstrates the complete closed-loop process of the method of the present invention, from sending to destruction, from interaction to the system, and from the terminal side to the cloud side, forming a secure, reliable, consistent, and flexibly adaptable multimedia self-destructing solution.

[0114] Example 2: Complete Implementation of Dual-Mode Interaction Between Partial Peeping and Panoramic Security

[0115] In an instant messaging application, user A sends a self-destructing image to user B. After user B clicks on the thumbnail on the receiving end, the system enters the secure preview stage and provides dual-mode interaction of partial peeping and full-view security.

[0116] (a) Partial Peeping Mode

[0117] When a user touches the security preview interface with a single finger, the system acquires the touch point coordinates in real time and generates a clear circular window centered on that point. The interior of the window is displayed with its original clarity, while the outer area maintains a Gaussian blur. The radius of the clear window can be dynamically adjusted based on touch pressure, swipe speed, or screen DPI, and the edges of the window use a gradient of transparency for a smooth transition. If the user switches from a single finger to two fingers, the system automatically transitions to the detection process of the panoramic security mode.

[0118] (II) Panoramic Security Mode

[0119] When a user needs to view the full image, the system requires them to simultaneously long-press the screen with two fingers. The system detects all touch points on the screen, and if exactly two touch points exist with a distance between them not less than a preset threshold, it is considered a valid "two-finger long-press" intent. Timing begins from the simultaneous pressing of two fingers. If the continuous pressing time reaches a preset transition duration T, a clear panoramic display is triggered. During the long press, the blur radius gradually decreases according to a non-linear transition curve. If any finger is released, a touch point moves off the screen, or an application switches to the background during the transition, the system immediately reverts to the safe preview state without consuming the viewing opportunity.

[0120] (III) Dynamic Configuration in the Cloud

[0121] The activation of the dual modes and all parameters can be dynamically configured via the cloud. The client retrieves the latest configuration upon startup or switching to the foreground, and the configuration takes effect after signature verification. Multiple configuration templates can be preset for different application scenarios, enabling flexible strategy adjustments.

[0122] (iv) Effect Verification

[0123] Tests showed that in partial viewing mode, bystanders could not identify the specific information on the screen, and user satisfaction reached 4.8 points; in panoramic security mode, only 3% of users successfully took photos with the help of auxiliary tools, effectively suppressing screen capture behavior.

[0124] Example 3: Adaptive Fuzzy Algorithm and Nonlinear Parameter Configuration Example

[0125] To clearly demonstrate the specific implementation effect of the adaptive Gaussian blur algorithm and nonlinear transition curve in this invention, this embodiment takes a typical mid-to-high-end smartphone as an example and provides a complete parameter configuration and calculation process.

[0126] (a) Equipment and Environment Configuration

[0127] The receiving device has a screen DPI of 420, which is considered a medium-to-high resolution model. The sending end presets the security level in the protocol field. (Medium security), viewing duration Second.

[0128] (II) Calculation of basic fuzzy radius

[0129] Based on the piecewise linear function of DPI:

[0130] When 320 ≤ DPI < 480, ;

[0131] Substitute DPI = 420:

[0132] ;

[0133] (III) Calculation of Content Complexity Factor

[0134] The system performs real-time analysis on the target image and calculates the gray-level co-occurrence matrix (GLCM): Contrast extraction: Contrast = 1250; Energy extraction: Energy = 0.32;

[0135] After normalization and weighting, the content complexity factor is obtained. A value greater than 1 indicates that the image content is relatively simple, making a higher level of blurring suitable for enhancing privacy protection.

[0136] (iv) Calculation of final fuzzy radius

[0137] Take weighting coefficients , Substitute into the formula:

[0138] .

[0139] (v) Determination of Gaussian kernel function parameters

[0140] Take the proportionality constant Then the standard deviation is: .

[0141] The two-dimensional Gaussian kernel function is:

[0142] .

[0143] (vi) Non-linear transition unlocking

[0144] Take the total transition time seconds, transition coefficient (The social scenario uses a fast-then-slow unlocking experience). During the user's long press, the blur radius changes over time:

[0145] ;

[0146] when seconds, (The blurriness decreases rapidly);

[0147] when seconds, (The change has slowed down);

[0148] when seconds, Enter clear display mode.

[0149] (vii) Effect verification

[0150] To verify the effectiveness of the adaptive fuzzy algorithm, comparative tests were conducted on different devices and with different content types. On devices with a screen DPI of 420, for simple content ( ) and complex content ( The tests were conducted separately, and the results are shown in Table 1.

[0151] Table 1. Comparison of fuzziness effect and user satisfaction under different content types.

[0152]

[0153] Test results show that the adaptive algorithm of this invention can automatically adjust the blur level according to the complexity of the content: a higher blur radius (53 px) is used for simple content to enhance privacy protection; for complex content, the blur radius is appropriately reduced (42 px) to retain more preview information for user confirmation. Compared with the traditional fixed radius scheme, user satisfaction is significantly improved.

[0154] For the nonlinear transition curve, this embodiment tested respectively. (Fast at first, then slow) (Linear) and The user accidental touch rate and satisfaction rate under the three configurations (slow first, then fast) are shown in Table 2.

[0155] Table 2 Comparison of User Experience under Different Transition Coefficients

[0156]

[0157] Tests show that in social scenarios, the use of It can provide the best user experience while ensuring a low rate of accidental touches.

[0158] This embodiment fully demonstrates the implementation effect of the adaptive Gaussian blur algorithm and nonlinear transition curve of the present invention through specific parameter configuration and test data, and verifies its superiority in security, user experience and cross-scene adaptability.

[0159] Example 4: Receipt Fault Tolerance Processing in Weak Network Scenarios

[0160] In practical applications, users may be in locations with unstable networks (such as subway tunnels, elevators, mountainous areas, or during network handover). When user B completes viewing and destruction in a weak network environment, the signature receipt generated by the client may not be uploaded to the server or sender in a timely manner. Therefore, this invention designs a fault-tolerant mechanism at the receiving end that includes persistent caching within the receiving end, exponential backoff retransmission, idempotent deduplication, and secure encryption, such as... Figure 6 As shown, this ensures eventual consistency even after prolonged periods of offline operation.

[0161] First, the receipt is synchronously written to a read-only destruction log and persisted to disk in a local queue within the same transaction. The queue is implemented using SQLite (with WAL enabled and secure_delete=ON) or an encrypted key-value store (such as a combination of SQLCipher / Keystore). The core fields of the queue list include:

[0162] `receipt_id` (UUID, a unique identifier for each receipt); `content_hash` (SHA-256 / BLAKE3 hash of the message body, corresponding one-to-one with the sender); `destroy_time_ms` (millisecond timestamp of destruction completion); `status` (queued / scheduled / retrying / delivered / expired); `retry_count`; `next_retry_at`; `last_err` (last failure code); `device_id_trunc` (device fingerprint truncation / salting value, used for audit alignment); `policy_digest` (summary of key policies for this session, used for compliance pattern comparison); `wm_nonce` (random factor for digital watermark session during the clearing phase); `payload_mac` (secondary MAC on the receipt payload to prevent local tampering). To avoid privacy leaks, sensitive fields only store necessary digest values ​​at the queue layer. The complete payload is encrypted and cached in real-time using the session key in a temporary area within the sandbox, and is destroyed along with the sandbox when the application is uninstalled.

[0163] When a network anomaly is detected (system network callback, DNS failure, TLS connection failure, cellular signal strength below the threshold, being in airplane mode / data saving mode, etc.), the upload status remains "not uploaded". A background keep-alive service (Android uses WorkManager / JobScheduler; iOS uses BGTaskScheduler / PushKit wake-up; desktop platforms use timers / system scheduled tasks) attempts to re-upload every 30 seconds, employing exponential backoff: 30s-60s-120s-240s-…, with a maximum of 10 minutes; each retry introduces 0–3s jitter to avoid amplification effects. If consecutive failures occur for more than 72 hours, the system will mark it as expired and discard the receipt, while simultaneously logging the receipt_id, content_hash, last error code, and last response digest into a read-only exception log for subsequent diagnostics. To improve bandwidth utilization in weak networks, retransmission supports two strategies at the queue level: prioritizing small loads and batch encapsulation. If an application moves to the foreground or the network transitions from "unavailable" to "available", an immediate retry is triggered.

[0164] To eliminate duplicates and out-of-order delivery, both the sender and the cloud use idempotent keys for receipt_id and content_hash:

[0165] If a processed receipt arrives again, only "Confirmed" will be returned, and the transaction will not be recorded again.

[0166] The cloud merges events for multi-device logins, using server time and signature validity as the determining factors to maintain a single "destroyed" final state.

[0167] If the receipt is found to be inconsistent with the session policy (policy_digest verification failed), the sender will record the receipt as "abnormal" and prompt the user to renegotiate the policy or report it to the administrator (enterprise mode).

[0168] To evaluate the retransmission performance under weak network conditions, this invention compared the acknowledgment arrival rate of traditional solutions with that of this solution under various network conditions. Tests were conducted under stable Wi-Fi, weak 4G signal, simulated 20% packet loss, high latency (RTT≈800ms), and cross-network handover conditions. The experiment was conducted under five conditions, and the results are shown in Table 3.

[0169] Table 3 Comparative test results of reply message delivery rates under different network environments

[0170]

[0171] The initial delivery time (the median from the completion of destruction to the display of "Destroyed" on the sending interface) was also evaluated, and the results are shown in Table 4.

[0172] Table 4. Comparison of median first delivery delay of receipt messages under different network environments

[0173]

[0174] This embodiment ensures that even in a weak network environment, the sending end can obtain the retransmission feedback information after the network is restored, thus guaranteeing the consistency of message processing and avoiding ambiguity between the two parties regarding the image processing status.

[0175] Example 5: Enterprise Compliance Audit Scenario

[0176] In industries such as finance, healthcare, and government, the "self-destructing message" function not only needs to meet user experience requirements but also needs to satisfy auditing and compliance requirements. This invention provides an optional "log auditing mode" for such scenarios.

[0177] When the enterprise administrator enables this function in the background, after the receiving-end terminal completes the destruction of the burn-after-reading pictures, it will not only generate a receipt feedback, but also generate an encrypted log record locally. This log includes: the unique content ID, the start time of viewing, the completion time of destruction, device fingerprint information, etc. The log is encrypted and stored using AES-256 and is regularly uploaded to the enterprise background server.

[0178] Compliance officers can decrypt and review these logs through the background management platform to trace the viewing situation of the content. This mechanism will neither affect the user's instant interaction experience nor meet the information traceability and auditing requirements of highly sensitive industries such as finance and healthcare.

[0179] Example Six: Cross-Platform Operation Mechanism

[0180] When designing the method of the present invention, cross-platform compatibility issues are considered. The client uses a platform-independent interface abstraction layer (API Layer) to unify the ability boundaries, and its module division is as Figure 7 shown: applyBlur (blur / text mask), watermark (clear phase bright / dark watermark), securePlay (secure video playback and minimum residency), clearMemory / clearGpuTexture / clearDecoder (memory / video memory / decoding buffer cleaning), secureErase (secure erase), startTimer (countdown), etc. The upper-layer logic only programs facing the API; the bottom layer is respectively mapped to:

[0181] Android: OpenGL ES / Vulkan + MediaCodec;

[0182] iOS / macOS: Metal / Core Image + AVFoundation;

[0183] Windows / Linux: Skia / OpenCV + FFmpeg (Vulkan / OpenCL / SSE / AVX when necessary);

[0184] HarmonyOS: ArkUI rendering and multimedia interface.

[0185] Preferably, the parameters are adaptively adjusted according to the device profile during runtime: on low-performance devices, the blur radius is automatically reduced by about 20%, and the transition duration is increased to 1.0 - 1.2 seconds to avoid frame drops; during the transition period (t < T), clear frames are strictly prohibited from entering the visible buffer, and when t ≥ T, clear decoding and watermark overlay are started for the same frame. Experiments show that the rendering delay on mainstream models is less than about 20ms, which can ensure smooth long-press transitions.

[0186] Example 7: Dynamic Configuration and Optimization of Parameters

[0187] To meet the needs of different application scenarios, this invention allows for dynamic adjustment of key parameters via a cloud interface. Operators can configure this in the backend.

[0188] Maximum viewing time (3–120 seconds);

[0189] Initial blur radius (20–50 pixels) and transition duration T;

[0190] Unlock methods (long press / swipe / double tap to confirm, etc.);

[0191] Retry strategy for uploading receipts (starting 30 seconds, index backoff limit ≤ 10 minutes);

[0192] Abnormal data discard timeout (24–120 hours);

[0193] Digital watermark switch and intensity, visible watermark drift rhythm, and hidden watermark embedding domain;

[0194] The minimum dwell time threshold for the video buffer (min_buffer_ms) and the maximum concurrent decoding capacity;

[0195] "Instant Destruction" triggers events on the white / black list (screenshot / screen recording / overlay / background switching, etc.).

[0196] The client automatically fetches the latest configuration upon startup or when switching to the foreground, expresses it in JSON format and transmits it via HTTPS. The configuration package is preferably signed and verified with ECDSA-P256 and then stored in a local read-only cache. If the network is unavailable, the previous version will be used.

[0197] For example, in social and entertainment scenarios, the duration can be set to 10 seconds, the blur radius to 20 pixels, and the watermark intensity reduced to improve the viewing experience; in enterprise office / financial scenarios, the duration can be set to 60 seconds and the blur radius to 40 pixels, the watermark intensity increased, and the upper limit for retransmission of receipts shortened, so that users can view complex documents for a longer period of time and enhance traceability.

[0198] The above description is merely a specific embodiment of this application, enabling those skilled in the art to understand or implement this application. Various modifications to these embodiments will be readily apparent to those skilled in the art, and the general principles defined herein may be implemented in other embodiments without departing from the spirit or scope of this application. Therefore, this application is not to be limited to the embodiments shown herein, but is to be accorded the widest scope consistent with the principles and novel features claimed herein.

[0199] As can be seen from the descriptions of the above embodiments, this invention proposes a method for processing self-destructing images based on fuzzy preview and long-press unlock. From sender identification, receiver state establishment, fuzzy preview, long-press unlock, clear display, timed and instant destruction, receipt generation and feedback, to cross-platform compatibility and dynamic configuration, it forms a complete and scalable technical system. Each link is independent yet closely connected, constituting a complete solution that can be implemented in practical applications.

[0200] It should be noted that the above embodiments are merely preferred embodiments of the present invention and do not constitute a limitation on the scope of protection of the present invention. Those skilled in the art can make various improvements or modifications to the specific steps, parameter ranges, module divisions, and system architecture of the present invention without departing from the principles and spirit of the present invention. For example, the fuzzy algorithm can not only employ Gaussian fuzzing, but also mean fuzzing, box fuzzing, or deep fuzzing algorithms based on convolutional neural networks; the unlocking method is not limited to long press, but can also employ multi-factor methods such as swipe confirmation, double-tap confirmation, and multi-finger touch; the destruction strategy is not limited to encryption key destruction, but can also be extended to methods such as encrypted erasure combined with hardware security modules.

[0201] While the specific embodiments of the present invention have been described above, they are not intended to limit the scope of protection of the present invention. Those skilled in the art should understand that various modifications or variations that can be made by those skilled in the art without creative effort based on the technical solutions of the present invention are still within the scope of protection of the present invention.

Claims

1. A method for processing self-destructing multimedia content based on fuzzy preview and long-press unlock, characterized in that, Includes the following processes: When a user selects target content, the sending end marks the content as self-destructing after reading. The mark includes at least filename modification and protocol field addition. The receiving end performs a security preview process during the initial display: Gaussian blur is applied to images, graphical rendering and masking are applied to text, and preview frame blurring or resolution reduction is applied to videos; the Gaussian blur applied to images has a blur radius that is dynamically adjusted non-linearly based on the screen pixel density (DPI) and content characteristics, specifically including: The basic blur radius is determined based on the screen DPI of the receiving device. And based on the content complexity factor of the target image and preset security level The final blur radius is calculated using the following formula: ； in and These are the weighting coefficients. , , The content complexity factor Quantization is achieved by calculating the contrast and energy values ​​of the image's gray-level co-occurrence matrix (GLCM). The value ranges from 0.5 to 1.5; the simpler the image content, the better. The larger the value, the higher the degree of ambiguity; the preset security level Defined by the sender in the message protocol field, the value ranges from 1 to 3, with higher values ​​indicating greater ambiguity; During the security preview phase, two security modes are provided: a partial peeping mode and a full-view security mode. In partial peeping mode, when a user touches the screen with a single finger, the coordinates of the touch point are obtained in real time, and a clear view is rendered centered on that point, while the surrounding area remains blurred, allowing the user to partially view the content. In full-view security mode, the system detects whether there are two touch points with a distance between them that is not less than a preset threshold. If the continuous pressing time reaches the preset transition duration, the system switches to full-view clear display. During the transition, if any finger is released, the touch point moves off the screen, the application is switched to the background, or the touch point jitter exceeds the tolerance range, the system immediately reverts to the security preview state without consuming viewing opportunities. Once the system enters clear display / playback mode, it starts a timer to display / play content according to the configured duration, and immediately destroys the content when the user releases their finger or the timer ends. After the destruction is complete, the receiving end generates a digitally signed receipt message and returns it to the sending end.

2. The method for processing self-destructing multimedia content based on fuzzy preview and long-press unlock as described in claim 1, characterized in that: After parsing the self-destructing message, the receiving end establishes a local access status record, which includes at least a unique content identifier, viewing status, receiving timestamp, and storage path. Multimedia content is stored locally in encrypted form using a session-dynamically generated symmetric encryption key. The key itself is protected by the platform's security interface and is bound to the device fingerprint, content hash, and policy parameters.

3. The method for processing self-destructing multimedia content based on fuzzy preview and long-press unlock as described in claim 1, characterized in that: The Gaussian blur applied to the image, in safe preview mode, achieves its blurring effect through a two-dimensional Gaussian kernel function, the standard deviation of which... With the final blur radius The relationship is ,in is a proportionality constant, ranging from 2.5 to 3.5; the formula for the two-dimensional Gaussian kernel function is: ； in, Indicates the convolution kernel in coordinates The weight value at the location; and These are the horizontal and vertical offsets relative to the center point of the convolution, in pixels. It is a natural exponential function.

4. The method for processing self-destructing multimedia content based on fuzzy preview and long-press unlock as described in claim 1, characterized in that: The basic fuzzy radius The DPI of the receiving device is determined piecewise linearly: when DPI ≥ 480, =45px; when hour, ;when DPI hour, Where 200 is the preset minimum DPI threshold, when DPI At 200, The blur radius is fixed at 20px; the blur transition time is 0.5–1 second; the blur radius gradually decreases to 0 using linear or exponential interpolation; and the user's interface displays the message "Release to destroy" during the unlocking process.

5. The method for processing self-destructing multimedia content based on fuzzy preview and long-press unlock as described in claim 1, characterized in that: In panoramic safety mode, the blur removal process follows a non-linear transition curve, with a blur radius... The change with time t is defined by the following formula: ； in, The duration of continuous pressure applied by the user. ; The preset total transition time ranges from 0.5 to 1.5 seconds. This is the transition coefficient, ranging from 0.8 to 2.0; when At that time, the fuzzy resolution process is fast at first and then slows down; when At that time, the fuzzy resolution process is slow at first and then fast; when When, it is a linear release; when hour, Enter clear display mode.

6. The method for processing self-destructing multimedia content based on fuzzy preview and long-press unlock as described in claim 1, characterized in that: The clear phase is overlaid with a dual-layer digital watermark (bright / dark) to suppress the shooting and circulation and enable source tracing; the bright watermark is a semi-transparent text or graphic that includes the device fingerprint, timestamp, and random sequence; the dark watermark is embedded through spatial domain LSB or frequency domain DCT / DFT perturbation and bound to the device fingerprint and random number to enable source tracing of the captured or forwarded images.

7. The method for processing self-destructing multimedia content based on fuzzy preview and long-press unlock as described in claim 1, characterized in that: The destruction mechanism includes the secure destruction of encryption keys, specifically: on the iOS platform, the Keychain interface is called to delete the key entry; on the Android platform, the Keystore / KeyMint interface is called to destroy the key; and on the Windows / Linux platform, the Cryptographic API or TPM-related interfaces are called to destroy the key. After the key is destroyed, even if the locally encrypted content file still exists, it cannot be decrypted and recovered.

8. The method for processing self-destructing multimedia content based on fuzzy preview and long-press unlock as described in claim 1, characterized in that: In weak network or network outage scenarios, the signed receipt message generated by the receiving end is written to the local queue cache, and the queue adopts a FIFO strategy; the client checks the network status every 30 seconds, and automatically retransmits the queue message after recovery; the retransmission mechanism adopts an exponential backoff strategy, with retry intervals of 30 seconds, 60 seconds, 120 seconds up to 10 minutes. If the transmission fails for more than 72 hours, it will be discarded and an exception log will be recorded.

9. A method for processing self-destructing multimedia content based on fuzzy preview and long-press unlock as described in claim 1, characterized in that: When the client starts up or switches to the foreground, it pulls the latest configuration parameters from the cloud. The configuration parameters include the fuzzy radius range, viewing duration, unlocking method, destruction strategy, watermark strength, and retransmission strategy. The client performs digital signature verification on the pulled configuration parameters, and stores them in the local read-only cache after the verification is successful. When the network is unavailable, the client uses the previous valid configuration parameters to perform subsequent processing.

Images