Digital image enhancement method and system based on endoscope

Abstract

The invention discloses a digital image enhancement method and system based on an endoscope, and aims to solve the problem of imaging limitation of a traditional endoscope in multi-focus positioning and targeted intervention. According to the digital image enhancement method based on the endoscope, endoscope image acquisition, wireless transmission, abnormal anchor point identification and controllable terminal targeted intervention operation can be carried out. The enhancement system for realizing the digital image enhancement method based on the endoscope identifies a suspected lesion point through image analysis, the control unit sends a release instruction to enable the endoscope to release the controllable terminal and establish communication, the scheduling system navigates the controllable terminal to an abnormal anchor point according to position information, and the abnormal anchor point is sent to the control unit. A focus image after digital image enhancement is collected and transmitted, and a judgment unit analyzes whether a target focus with an intervention value exists or not. According to the method, the digital image in the human body cavity can be enhanced, and the focus recognition accuracy and intervention efficiency are improved.

Description

Technical Field

[0001] The present invention relates to the technical field of image processing and intelligent diagnosis and treatment control, and particularly to a digital image enhancement method and system based on an endoscope. Background Art

[0002] As an important tool for digestive tract examination and treatment, endoscopes are widely used in the early screening, biopsy and interventional operations of digestive diseases. Although traditional electronic bronchoscopes have the ability to acquire high-resolution images, their efficiency in acquiring digital images in complex anatomical regions (such as tissues like epiglottis, piriform fossa, tracheal ring, carina, etc.) is relatively low, and the distortion degree of the actual images is relatively high. In this case, medical diagnosis relies on the subjective experience judgment of medical operators. In the case of high image distortion, existing endoscopes often need to replace instruments or repeat endoscope insertion, resulting in low operation efficiency, increased patient pain, and easy omission of occult lesions, which is not conducive to the development needs of refined diagnosis and treatment.

[0003] In order to further improve the efficiency of digital image acquisition of endoscopes, related endoscope digital image enhancement algorithms have been disclosed in the prior art. This digital image enhancement algorithm realizes the repair and optimization of the lesion images collected by the endoscope from the perspective of image processing. In specific diagnosis and treatment scenarios, most of the existing methods rely on image acquisition from a single perspective, and most of the related digital image enhancement algorithms in the prior art are post-processing processes of images, lacking dynamic linkage with the image acquisition process, unable to provide real-time feedback and optimize the image acquisition path, thus affecting the diagnosis and treatment response efficiency. As a result, when clinical doctors observe images, they often need to repeatedly check suspicious areas in multiple image frames, the subjective interpretation burden of the images is relatively heavy, and it is easy to miss important lesions due to lagged operations. Therefore, the prior art urgently needs to be further improved. Summary of the Invention

[0004] In view of the above problems, the present invention provides a digital image enhancement method based on an endoscope. In this enhancement method, the endoscope serves as a front-end recognition and image acquisition device, collects images of the lesion area, judges potential abnormal areas through image processing, identifies and marks one or more abnormal anchor points, generates corresponding spatial position information and recognized image features, and sends them to a controllable terminal. The controllable terminal can be a device with propulsion and autonomous response capabilities. After receiving the abnormal anchor point position information, it calculates the navigation path, moves to the preset target lesion area, and executes various programmable intervention actions. Further, the present invention also provides an enhancement system for implementing the digital image enhancement method based on an endoscope.

[0005] The invention object of the present application can be achieved by the following technical means:

[0006] A digital image enhancement method based on an endoscope, comprising the following steps:

[0007] Step 1: The endoscope acquires n groups of lesion area images p. The n groups of lesion area images p are encoded to generate an original video stream, and the original video stream is transmitted to the wireless transmission unit;

[0008] Step 2: The wireless transmission unit sends the original video stream to the medical terminal, and the medical terminal decodes the original video stream and transmits it to the control unit and the display unit;

[0009] Step 3: If the display unit inputs a first communication instruction to the control unit, go to Step 4; otherwise, return to Step 1;

[0010] Step 4: The control unit sends decision information to the endoscope and records at least one abnormal anchor point. The endoscope releases the controllable terminal and establishes a channel connection between the endoscope and the controllable terminal;

[0011] Step 5: The endoscope extracts the position information of the abnormal anchor point and sends the position information to the controllable terminal;

[0012] Step 6: The scheduling system adjusts the position of the controllable terminal and moves it to the abnormal anchor point. The controllable terminal acquires m groups of lesion area images q. The m groups of lesion area images q are encoded to generate an enhanced video stream, and the enhanced video stream is transmitted to the wireless transmission unit;

[0013] Step 7: The wireless transmission unit sends the enhanced video stream to the medical terminal, and the medical terminal sends the enhanced video stream to the wireless transmission unit and the judgment unit;

[0014] Step 8: The determination unit receives the enhanced video stream and extracts the x-frame enhanced image Y of the enhanced video stream x , and traverses the x-frame enhanced image Y x ;

[0015] Step 9: If there is at least one frame of enhanced image Y i identified as the target lesion, go to Step 10; otherwise, return to Step 5, i = 1, 2,..., x;

[0016] Step 10: The wireless transmission unit transmits the enhanced image Y i to the display unit.

[0017] In the present invention, the first communication instruction is a start intervention command input by the user through the display unit interaction interface, and the control unit starts the recognition of the abnormal anchor point and completes the release operation of the controllable terminal based on this communication instruction.

[0018] In the present invention, in Step 4, the decision information includes a controllable terminal release instruction and channel parameters required for establishing a communication channel between the endoscope and the controllable terminal.

[0019] In the present invention, the abnormal anchor point is a suspected pathological feature marker point identified by an image analysis module in the lesion area image p collected by an endoscope.

[0020] In the present invention, the feature marker point includes the texture, color, edge, and contrast abnormal features of the lesion area image p, and the comprehensive determination basis is the abnormal scoring function S(x, y), , where: is the image pixel coordinate; represents the gradient amplitude of this point, reflecting the edge change intensity; represents the texture feature value extracted by local binary pattern, is the texture mean value of the whole image; represents the color vector of the image in the CIELab color space, is the regional color mean value; represents the contrast index calculated based on the gray-level co-occurrence matrix, is its global mean value; are the weighting coefficients of each feature, satisfying , i = 1, 2,..., 4; when (where is a preset threshold), it is determined that this point is an abnormal anchor point and its coordinate information is recorded.

[0021] In the present invention, the two-dimensional image coordinates (x, y) of the abnormal anchor point in the endoscope image p are extracted, and the timestamp K of the first communication instruction is read t , through the timestamp K t to match the position information collected by the attitude sensor of the controllable terminal at this moment.

[0022] In the present invention, the position information includes the direction angle, depth value, and displacement vector of the endoscope, and the position information is used to control the scheduling system to adjust the position of the controllable terminal.

[0023] In the present invention, the target lesion refers to the target pathological area collected by the controllable terminal in the x-frame enhanced image Y x and determined by the judgment unit to have intervention value.

[0024] In the present invention, the target pathological area includes at least two image feature matching results:

[0025] a) The degree of edge contour mutation exceeds the set threshold , that is, the gradient amplitude of the lesion edge satisfies: ;

[0026] b) The local color deviates from the reference value of the central area by more than the color difference threshold , that is: ;

[0027] c) The gray - scale contrast and texture structure of the region are more similar to the depth model of the target lesion than the comparison threshold , by the similarity function Satisfy: ;

[0028] Where: is the gray - scale value of the pixel point in image q; is the color vector in the CIELab color space; is the candidate region image patch, is the depth model of the target lesion; represents the similarity function.

[0029] An enhancement system for implementing an endoscope - based digital image enhancement method, including an endoscope, a controllable terminal, a scheduling system, a wireless transmission unit, a judgment unit, a medical terminal, a display unit, and a control unit. Among them,

[0030] The endoscope is used to collect the lesion region image p, encode the image to generate the original video stream, release the controllable terminal according to the decision information sent by the control unit, and establish a channel connection with the controllable terminal at the same time;

[0031] The controllable terminal is used to move to the abnormal anchor point position under the action of the scheduling system after receiving the abnormal anchor point position information sent by the endoscope, collect the lesion region image q, and generate an enhanced video stream;

[0032] The scheduling system is used to perform directional control and spatial attitude adjustment on the controllable terminal according to the azimuth angle, depth value, and displacement vector in the position information;

[0033] The wireless transmission unit is used to receive n groups of enhanced video streams and transmit them to the medical terminal respectively, and establish a data communication channel between the endoscope and the controllable terminal;

[0034] The judgment unit is used to perform image feature analysis on x frames of enhanced images Y in the enhanced video stream x and judge whether there is a target lesion based on edge gradient, color offset, and texture similarity;

[0035] The medical terminal is used to receive and decode the video stream sent by the wireless transmission unit, and transmit the decoded image data to the control unit and the display unit respectively;

[0036] The display unit is used to display images p and q, for the user to interactively input the first communication instruction, and is used to receive and display the finally back - transmitted enhanced video stream image;

[0037] The control unit is used to respond to the first communication instruction input by the display unit, start the image recognition, coordinate extraction, position information calculation, controllable terminal release and task scheduling of the abnormal anchor point, and at the same time coordinate and schedule the system to complete the path control of the controllable terminal, and call the judgment unit to identify the target lesion.

[0038] Implementing a digital image enhancement method and system based on an endoscope according to the present invention has the beneficial effects that: the technical solution disclosed by the present invention constructs a task separation mechanism between the main control endoscope and the responsive controllable terminal, significantly improving the recognition efficiency and intervention accuracy of suspicious lesions in a complex human body cavity environment, and breaking through the technical limitation that it is difficult for traditional single-device operation to take into account both recognition and intervention. Further, by extracting the coordinates of abnormal anchor points in the image and combining the attitude sensing information and the time synchronization mechanism, the system stability, operation safety and remote interaction ability are significantly improved. Compared with traditional electronic endoscopes, it has stronger operability and plays a significant role in reducing the diagnosis and treatment time. Brief Description of the Drawings

[0039] Figure 1 is a flowchart of a digital image enhancement method based on an endoscope according to the present invention;

[0040] Figure 2 is a flowchart of the spatial position information extraction process of the digital image enhancement method based on an endoscope;

[0041] Figure 3 is a hardware block diagram of an enhancement system for implementing a digital image enhancement method based on an endoscope. Detailed Embodiments

[0042] The following further describes the present invention in conjunction with the drawings and embodiments for those skilled in the art to understand and implement the present invention. It should be understood, however, that these embodiments are only used to illustrate the present invention and not to limit the present invention. Without departing from the spirit and essence of the present invention, those skilled in the art can make various equivalent deformations and substitutions of the present invention, which should all be covered by the protection scope of the present invention.

[0043] Embodiment 1

[0044] A digital image enhancement method based on an endoscope, referring to Figure 1 , includes the following steps:

[0045] Step 1: The endoscope acquires n groups of lesion area images p. The n groups of lesion area images p are encoded to generate an original video stream, which is transmitted to the wireless transmission unit. In this embodiment, the image resolution of the images p acquired by the endoscope is preferably 1920×1080 pixels, and the frame rate is 30 frames per second. After being processed by the built-in video encoding module, the images are compressed using the H.265 encoding method to generate n groups of independent video streams, and each group of video streams corresponds to a different image time period or focal plane area.

[0046] Step 2: The wireless transmission unit sends the original video stream to the medical terminal, and the medical terminal decodes the original video stream and transmits it to the control unit and the display unit. In this embodiment, the original video stream carries frame sequence tags and timestamp information during the transmission process, which is used for the medical terminal to perform time alignment on the image data during the decoding stage.

[0047] Furthermore, a decoding scheduling module is provided inside the medical terminal, which splits the original video stream into two types of data paths, namely the control processing stream and the display presentation stream, according to the data priority: The control processing stream includes key frames, image marking information, and metadata, which are preferentially decoded and then pushed to the control unit in real time; the display presentation stream contains the full-frame image sequence, which is cached in order and synchronously output to the display unit for image playback and user interaction.

[0048] Step 3: If the display unit inputs a first communication instruction to the control unit, then go to Step 4; otherwise, return to Step 1. In this embodiment, the first communication instruction is a start intervention command input by the user through the display unit's interaction interface, and the control unit starts the recognition of abnormal anchor points and completes the release operation of the controllable terminal based on this communication instruction.

[0049] Step 4: The control unit sends decision information to the endoscope and records at least one abnormal anchor point. The endoscope releases the controllable terminal and establishes a channel connection between the endoscope and the controllable terminal. In this embodiment, the decision information includes a controllable terminal release instruction and channel parameters required for the endoscope and the controllable terminal to establish a communication channel.

[0050] Step 5: The endoscope extracts the position information of the abnormal anchor point and sends the position information to the controllable terminal. The abnormal anchor point is a suspected pathological feature marker point identified by the image analysis module in the lesion area image p acquired by the endoscope.

[0051] In this embodiment, the feature marker points include the texture, color, edges, and contrast abnormal features of the lesion area image p, and the comprehensive determination basis is the abnormal scoring function S(x,y), , where: is the image pixel coordinate; represents the gradient amplitude of this point, reflecting the edge change intensity; Represents the texture feature value extracted by local binary pattern, which is the texture mean value of the whole image; Represents the color vector of the image in the CIELab color space, which is the regional color mean value; Represents the contrast index calculated based on the gray-level co-occurrence matrix, and its global mean value is; is the weighted coefficient of each feature, satisfying , i = 1, 2,..., 4; when (where is a preset threshold), determine that this point is an abnormal anchor point and record its coordinate information.

[0052] In this embodiment, extract the two-dimensional image coordinates (x, y) of the abnormal anchor point in the endoscopic image p, and read the timestamp K of the first communication instruction t , and match the position information collected by the attitude sensor of the controllable terminal at this moment through the timestamp K t . The position information includes the direction angle, depth value, and displacement vector of the endoscope, and the position information is used to control the scheduling system to adjust the position of the controllable terminal.

[0053] Step 6: The scheduling system adjusts the position of the controllable terminal and moves it to the abnormal anchor point. The controllable terminal collects m sets of lesion area images q. After encoding, the m sets of lesion area images q generate an enhanced video stream, and the enhanced video stream is transmitted to the wireless transmission unit.

[0054] Step 7: The wireless transmission unit sends the enhanced video stream to the medical terminal, and the medical terminal sends the enhanced video stream to the wireless transmission unit and the judgment unit.

[0055] Step 8: The judgment unit receives the enhanced video stream and extracts the x-frame enhanced image Y of the enhanced video stream x , and traverse the x-frame enhanced image Y x .

[0056] Step 9: If there is at least one frame of enhanced image Y i identified as the target lesion, go to Step 10; otherwise, return to Step 5, i = 1, 2,..., x. The target lesion refers to the target pathological area that is collected by the controllable terminal and determined by the judgment unit to have intervention value in the x-frame enhanced image Y x .

[0057] In this embodiment, the target pathological area includes at least two image feature matching results:

[0058] a) The degree of edge contour mutation exceeds the set threshold , that is, the gradient amplitude of the lesion edge satisfies: ;

[0059] b) The local color deviates from the reference value of the central region exceeding the color difference threshold , that is: ;

[0060] c) The gray-scale contrast and texture structure of the region are more similar to the depth model of the target lesion than the comparison threshold , which is satisfied by the similarity function : ;

[0061] where: is the gray value of the pixel point in image q; is the color vector in the CIELab color space; is the candidate region image block, is the depth model of the target lesion; represents the similarity function.

[0062] Step 10: The wireless transmission unit transmits the enhanced image Y i to the display unit.

[0063] Embodiment 2:

[0064] To further enhance the clinical visualization effect and operation intuitiveness of digital images, in a specific implementation of the technical solution disclosed in this application, an endoscopic digital image enhancement method based on mixed reality technology is provided. This embodiment fuses traditional two-dimensional images with three-dimensional spatial information to achieve holographic visual annotation of the target lesion, and based on real-time spatial superposition guidance, effectively improves the spatial perception and intervention accuracy of doctors during endoscopic operations.

[0065] In step 5, after the endoscope identifies and marks the abnormal anchor points, it needs to extract their spatial position information and send it to the controllable terminal for it to achieve the moving target in step 6. Referring to Figure 2 , the extraction process of this spatial position information includes the following steps:

[0066] Step 201: The endoscope processes the lesion area image p through the image analysis module, identifies the image points with abnormal image features, and records their two-dimensional image coordinates in image p, denoted as (x, y);

[0067] Step 202: After the user inputs the first communication instruction through the display unit, the control unit records the current timestamp K t , and this timestamp is used as a reference for synchronizing and matching the image p and the attitude sensor data.

[0068] Step 203: According to the timestamp K t, match the spatial attitude data of the endoscope recorded by the controllable terminal attitude sensor at time t, and generate a homogeneous transformation matrix: , where represents the rotation matrix formed by the direction angles; represents the displacement vector of the endoscope in the body cavity.

[0069] Step 204: Assume that the endoscope imaging system satisfies the pinhole camera model. If the pixel coordinates of the abnormal anchor point in the image p are (x, y), then the camera internal parameter matrix is: , where are the focal lengths in the horizontal and vertical directions, are the principal point coordinates. Then the normalized light direction vector corresponding to this pixel point can be expressed as: .

[0070] Step 205: Denote the depth value of the abnormal anchor point as , then the spatial position of this anchor point in the local coordinate system of the endoscope is: , that is, after magnifying to the real depth according to the unit normalization direction to obtain the three-dimensional point coordinates. Then the reference coordinate of the abnormal anchor point in the three-dimensional space of the human body cavity can be expressed as: , where is the spatial position of the abnormal anchor point under the endoscope view, that is, the spatial position information of this point, which is used for the target setting of the movement path.

[0071] In this embodiment, the homogeneous transformation matrix is used to realize the mapping between the image coordinate system and the three-dimensional space coordinate system. It should be understood that this technical solution realizes the precise positioning of image information in space through a clear coordinate transformation relationship, which is one of the technical features of the present invention. Although there are also coordinate mapping methods based on homogeneous matrices in the prior art, the matrix construction method defined in the present invention and its linkage relationship with the control unit have clear structural limitations and functional coordination, which are different from the existing conventional processing methods.

[0072] Embodiment Three

[0073] This embodiment details a specific method for target lesion determination, which is used to implement the digital image enhancement method based on the endoscope. After the controllable terminal reaches the abnormal anchor point position in step 6, it acquires the lesion area image q. This image q is encoded to generate an enhanced video stream, which is transmitted to the judgment unit by the wireless transmission unit. The task of the judgment unit is to analyze the image features in the image q and identify whether there is a target lesion according to the set judgment criteria. If so, execute step 9; otherwise, return to step 5 for anchor point adjustment.

[0074] In step 9, the target lesion refers to in the x-frame enhanced image Y xThe target pathological region collected by the controllable terminal and determined by the judgment unit to have intervention value. The judgment unit performs inter-frame segmentation processing on the image q, intercepts m frames of images, analyzes the image features such as spatial structure, color distribution, and gray texture frame by frame, and matches them with the preset model.

[0075] In this embodiment, the determination of the target lesion does not depend on a single image feature, but rather fuses three key indicators: edge mutation, color offset, and texture similarity. The determination process of the key indicators is based on logical matching of the quantization threshold and the image similarity function, supporting the deployment of the algorithm module.

[0076] Determination condition 1: For any pixel point (x, y) in the abnormal anchor point in the present invention, calculate the gray gradient amplitude of this pixel point: . If there is an area that satisfies , then this area is determined to have the characteristic of abnormal edge contour mutation. Among them, is the set threshold, where I q is the component channel of the lesion area image q of the image.

[0077] Determination condition 2: Convert the image q to the CIELab color space and perform color vector analysis on each pixel point. Define the color vector of this point as: , with the color reference mean of the central area being , define the color difference as: . If it satisfies: , then it is considered that the color offset of this area is significant and has the characteristic of color abnormality. Among them, is the color difference threshold, L q (x, y) is the brightness component of the lesion area image q, a q (x, y) is the color component of the lesion area image q from green to red, b q (x, y) is the color component of the lesion area image q from blue to yellow.

[0078] Determination condition 3: Let a certain candidate image area be , the depth model or template image of the target lesion be , use the gray-level co-occurrence matrix to extract the local texture structure and the reference texture , and calculate the similarity through the structural similarity function or the cosine similarity function : . If it satisfies: , then it indicates that the candidate area is highly similar to the target lesion model and has the typical pathological structure characteristics, where is the comparison threshold.

[0079] In this embodiment, for the first determination condition, considering that under complex conditions, the grayscale gradient amplitude cannot be accurately calculated, the following supplementary first determination condition is based on: the average gradient values of the pixel points corresponding to the abnormal anchor points within the 3×3 or 5×5 neighborhood range all satisfy the threshold condition, and the edge of the region is continuous, that is, the contour of the abnormal region needs to form a closed or semi-closed shape to avoid misidentifying isolated high-gradient noise points as target lesions.

[0080] Embodiment Four

[0081] An enhancement system for implementing an endoscopic-based digital image enhancement method, referring to Figure 3 , includes an endoscope, a controllable terminal, a scheduling system, a wireless transmission unit, a judgment unit, a medical terminal, a display unit, and a control unit.

[0082] The endoscope is used to collect the image p of the lesion area, encode the image to generate the original video stream, and release the controllable terminal according to the decision information sent by the control unit, and at the same time establish a channel connection with the controllable terminal. In this embodiment, the endoscope is a flexible insertion device with high-definition imaging and video coding modules. The coding module realizes multi-channel video stream output based on the H.265 protocol, and establishes a low-latency data link with the controllable terminal through preset channel parameters.

[0083] The controllable terminal is used to move to the abnormal anchor point position under the action of the scheduling system after receiving the abnormal anchor point position information sent by the endoscope, and collect the image q of the lesion area to generate an enhanced video stream. In this embodiment, the controllable terminal can complete the movement under the control of the scheduling system and encode and output the enhanced video stream at a fixed frame rate.

[0084] The scheduling system is used to perform directional control and spatial attitude adjustment on the controllable terminal according to the azimuth angle, depth value, and displacement vector in the position information. The scheduling system is composed of an electromagnetic coil array, and preferably the magnetic control response frequency range is 75 - 100Hz.

[0085] The wireless transmission unit is used to receive n groups of enhanced video streams and transmit them to the medical terminal respectively, and establish a data communication channel between the endoscope and the controllable terminal. Preferably, the wireless transmission unit supports the Wi-Fi and Bluetooth dual-channel communication modes, has an automatic channel switching and anti-interference mechanism, and can ensure the high-bandwidth and low-latency data synchronization requirements.

[0086] The judgment unit is used to perform image feature analysis on the x-frame enhanced image Y in the enhanced video stream x to determine whether there is a target lesion based on edge gradient, color offset, and texture similarity. In this embodiment, the judgment unit is integrated in the medical terminal, and performs inter-frame processing on the lesion area image q through a convolutional neural network. It should be understood that a convolutional neural network model and an edge gradient threshold are pre-stored in the judgment unit Color difference threshold Similarity threshold 。

[0087] The medical terminal is used to receive and decode the video stream sent by the wireless transmission unit, and transmit the decoded image data to the control unit and the display unit respectively. In this embodiment, the medical terminal can be any device with the ability to input control signals.

[0088] The display unit is used to display image p and image q for the user to interactively input the first communication instruction, and is used to receive the enhanced video stream image for final display.

[0089] The control unit is used to respond to the first communication instruction input by the display unit, start the image recognition, coordinate extraction, position information calculation of abnormal anchor points, release of controllable terminals and task scheduling, coordinate and schedule the system to complete the path control of the controllable terminal, and call the judgment unit to identify the target lesion.

[0090] The above are only the preferred embodiments of the present invention and are not intended to limit the present invention. Any modifications, equivalent replacements, improvements, etc. made within the spirit and principle of the present invention shall be included in the protection scope of the present invention.

Claims

1. A digital image enhancement method based on an endoscope, characterized in that, It includes the following steps: Step 1: The endoscope acquires n groups of lesion area images p. The n groups of lesion area images p are encoded to generate an original video stream, and the original video stream is transmitted to the wireless transmission unit; Step 2: The wireless transmission unit sends the original video stream to the medical terminal, and the medical terminal decodes the original video stream and transmits it to the control unit and the display unit; Step 3: If the display unit inputs a first communication instruction to the control unit, then go to Step 4; Otherwise, return to Step 1; Step 4: The control unit issues decision information to the endoscope, and records at least one abnormal anchor point. The endoscope releases the controllable terminal and establishes a channel connection between the endoscope and the controllable terminal; Step 5: The endoscope extracts the position information of the abnormal anchor point and sends the position information to the controllable terminal; Step 6: The scheduling system adjusts the position of the controllable terminal and moves it to the abnormal anchor point. The controllable terminal acquires m groups of lesion area images q. The m groups of lesion area images q are encoded to generate an enhanced video stream, and the enhanced video stream is transmitted to the wireless transmission unit; Step 7: The wireless transmission unit sends the enhanced video stream to the medical terminal, and the medical terminal sends the enhanced video stream to the wireless transmission unit and the judgment unit; Step 8: The determination unit receives the enhanced video stream and extracts the x-frame enhanced image Y of the enhanced video stream x , and traverses the x-frame enhanced image Y x ; Step 9: If there is at least one frame of enhanced image Y i identified as a target lesion, then proceed to Step 10; otherwise, return to Step 5, where i = 1, 2,..., x; Step 10: The wireless transmission unit transmits the enhanced image Y i to the display unit.

2. The endoscopic-based digital image enhancement method according to claim 1, wherein The first communication instruction is a start intervention command input by the user through the display unit's interaction interface. The control unit starts the recognition of the abnormal anchor point and completes the release operation of the controllable terminal based on this communication instruction.

3. The endoscopic-based digital image enhancement method according to claim 1, wherein In Step 4, the decision information includes a controllable terminal release instruction and channel parameters required for establishing a communication channel between the endoscope and the controllable terminal.

4. The endoscopic-based digital image enhancement method according to claim 1, wherein The abnormal anchor point is a suspected pathological feature marker point identified by the image analysis module in the lesion area image p acquired by the endoscope.

5. The endoscopic-based digital image enhancement method according to claim 4, characterized in that The feature marker points include the texture, color, edges, and contrast anomaly features of the lesion area image p, and the comprehensive determination basis is the anomaly scoring function S(x, y), , where: are the image pixel coordinates; represents the gradient magnitude of the point, reflecting the edge change intensity; represents the texture feature value extracted by local binary pattern, is the mean texture of the whole image; represents the color vector of the image in the CIELab color space, is the regional color mean; represents the contrast index calculated based on the gray-level co-occurrence matrix, is its global mean; are the weighting coefficients of each feature, satisfying , i = 1, 2,..., 4; when , determine that the point is an abnormal anchor point and record its coordinate information, where is the preset threshold.

6. The endoscopic-based digital image enhancement method according to claim 1, wherein Extract the two-dimensional image coordinates (x, y) of the abnormal anchor points in the endoscopic image p, and read the timestamp K of the first communication instruction. t , through the timestamp K t Match the position information collected by the controllable terminal attitude sensor at this moment.

7. The endoscopic-based digital image enhancement method according to claim 1, wherein The position information includes the direction angle, depth value, and displacement vector of the endoscope. The position information is used to control the scheduling system to adjust the position of the controllable terminal.

8. The endoscopic-based digital image enhancement method according to claim 1, characterized in that, The target lesion refers to the target pathological area collected by the controllable terminal in the x-frame enhanced image Y x and determined by the judgment unit to have intervention value.

9. The endoscopic-based digital image enhancement method according to claim 8, wherein The target pathological area includes at least two image feature matching results: a) The degree of mutation of the edge contour exceeds the set threshold , that is, the gradient amplitude of the lesion edge satisfies: ; b) The local color deviates from the reference value of the central area exceeds the color difference threshold , that is: ; c) The similarity of the regional gray contrast and texture structure to the depth model of the target lesion is higher than the comparison threshold , which is satisfied by the similarity function : ; Wherein: is the gray value of the pixel point in image q; is the color vector in the CIELab color space; is the candidate region image patch, is the depth model of the target lesion; represents the similarity function.

10. An enhancement system for implementing the endoscopic-based digital image enhancement method as claimed in claim 1, characterized in that, It includes an endoscope, a controllable terminal, a scheduling system, a wireless transmission unit, a judgment unit, a medical terminal, a display unit, and a control unit. Among them, The endoscope is used to acquire the lesion area image p, encode the image to generate an original video stream, release the controllable terminal according to the decision information sent by the control unit, and establish a channel connection with the controllable terminal at the same time; The controllable terminal is used to move to the abnormal anchor point position under the action of the scheduling system after receiving the abnormal anchor point position information sent by the endoscope, and acquire the lesion area image q to generate an enhanced video stream; The scheduling system is used to perform directional control and spatial attitude adjustment on the controllable terminal according to the direction angle, depth value, and displacement vector in the position information; The wireless transmission unit is used to receive the n groups and the enhanced video stream and transmit them to the medical terminal respectively, and establish a data communication channel between the endoscope and the controllable terminal; The judgment unit is used to perform image feature analysis on the x enhanced images Y in the enhanced video stream x to determine whether there is a target lesion based on edge gradient, color offset, and texture similarity; The medical terminal is used to receive and decode the video stream sent by the wireless transmission unit, and transmit the decoded image data to the control unit and the display unit respectively; The display unit is used to display the image p and the image q, for the user to interactively input the first communication instruction, and is used to receive the finally back-transmitted enhanced video stream image for display; The control unit is used to respond to the first communication instruction input by the display unit, start the image recognition, coordinate extraction, position information calculation, controllable terminal release and task scheduling of the abnormal anchor point, and at the same time coordinate and dispatch the system to complete the path control of the controllable terminal, and call the judgment unit to identify the target lesion.

Images