A digital image enhancement method and system based on endoscope

Abstract

The present invention discloses an endoscope-based digital image enhancement method and system, which aims to solve the imaging limitations of traditional endoscopes in multi-lesion positioning and targeted intervention. The endoscope-based digital image enhancement method can perform endoscopic image acquisition, wireless transmission, abnormal anchor point identification and controllable terminal targeted intervention operations. The enhancement system for implementing the endoscope-based digital image enhancement method identifies suspected lesions through image analysis, the control unit issues a release instruction, causes the endoscope to release the controllable terminal and establish communication, the scheduling system navigates the controllable terminal to the abnormal anchor point according to the position information, acquires and transmits the digitally enhanced lesion image, and the judgment unit analyzes whether there is a target lesion with intervention value. This method can enhance the digital image in the human body cavity and improve the accuracy of lesion identification and intervention efficiency.

Description

Technical Field

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

[0002] Endoscopes are important tools for digestive tract examination and treatment and are widely used for early screening, biopsy, and interventional procedures for digestive diseases. Although traditional electronic bronchoscopes have the ability to acquire high-resolution images, their efficiency in acquiring digital images of complex anatomical areas (such as the epiglottis, piriform whorls, tracheal rings, carina, and other tissues) is low, and the actual image distortion is high. In this case, medical diagnosis relies on the subjective experience and judgment of medical operators. In cases of high image distortion, existing endoscopes often require instrument replacement or repeated insertion, resulting in low operational efficiency, increased patient pain, and the easy omission of hidden lesions, which is not conducive to the development of refined diagnosis and treatment.

[0003] In order to further improve the efficiency of digital image acquisition by endoscopes, the prior art discloses a related endoscopic digital image enhancement algorithm, which realizes the repair and optimization of lesion images collected by endoscopes from the perspective of image processing. In specific diagnosis and treatment scenarios, most of the existing methods rely on image acquisition under a single perspective, and the digital image enhancement algorithms related to this field in the prior art are mostly image post-processing processes, lacking dynamic linkage with the image acquisition process, and unable to provide real-time feedback and optimize the image acquisition path, thereby affecting the efficiency of diagnosis and treatment response. As a result, when observing images, clinicians often need to repeatedly check suspicious areas in multiple image frames, which places a heavy burden on the subjective interpretation of images and is prone to missing important lesions due to operational delays. Therefore, the prior art urgently needs further improvement. Summary of the Invention

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

[0005] The invention objectives of this application can be achieved through the following technical means:

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

[0007] Step 1: The endoscope collects n sets of lesion area images p, and the n sets of lesion area images p are encoded to generate original video streams, which are 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 display unit;

[0009] Step 3: If the display unit inputs the first communication command to the control unit, proceed to step 4; otherwise, return to step 1;

[0010] Step 4: The control unit sends a decision message 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 collects m groups of lesion area images q. The m groups of lesion area images q are encoded to generate enhanced video streams, which are 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 decision unit receives the enhanced video stream and extracts the x-frame enhanced image Y of the enhanced video stream. x , traverse the x frame enhanced image Y x ;

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

[0016] Step 10: The wireless transmission unit will enhance the image Y i Transmitted to the display unit.

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

[0018] In the present invention, in step 4, the decision information includes the controllable terminal release instruction and the channel parameters required for the endoscope to establish a communication channel with 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 through an endoscope.

[0020] In the present invention, the characteristic markers include the texture, color, edge and contrast abnormality features of the lesion area image p, and the comprehensive judgment basis is the abnormality scoring function S(x, y). ,in: is the image pixel coordinate; Indicates the gradient amplitude of the point, reflecting the intensity of edge change; represents the texture feature value extracted by the local binary pattern, is the texture mean of the entire 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; is the weighted coefficient of each feature, satisfying , i=1,2,...,4; when (in is the preset threshold), the point is determined to be 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 endoscopic image p are extracted, and the timestamp K of the first communication instruction is read. t , through the timestamp K t Match the position information collected by the controllable terminal posture sensor at that 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 enhanced image Y in frame x. x The target pathological area is collected by the controllable terminal 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 edge profile mutation exceeds the set threshold , that is, the gradient amplitude of the lesion edge satisfies: ;

[0026] b) Local color deviates from the central area reference value Exceeding the color difference threshold ,Right now: ;

[0027] c) The similarity between the regional grayscale contrast and texture structure and the depth model of the target lesion is higher than the contrast threshold , by the similarity function satisfy: ;

[0028] in: is the grayscale value of the pixel in image q; is the color vector in CIELab color space; is the candidate region image block, 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 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, wherein:

[0030] The endoscope is used to collect images p of the lesion area, encode the images to generate original video streams, release the controllable terminal according to the decision information issued by the control unit, and establish a channel connection with the controllable terminal;

[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, and collect the image q of the lesion area to generate an enhanced video stream;

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

[0033] a wireless transmission unit for receiving n groups and enhanced video streams and transmitting them to the medical terminal respectively, and establishing a data communication channel between the endoscope and the controllable terminal;

[0034] The judgment unit is used to enhance the image Y of frame x in the enhanced video stream. x Perform image feature analysis to determine whether a target lesion exists based on edge gradient, color shift, 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 the image p and the image q for the user to interactively input the first communication instruction, and to receive the enhanced video stream image display that is finally transmitted back;

[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 and controllable terminal release and task scheduling of the abnormal anchor point, and coordinate the scheduling system to complete the controllable terminal path control and call the judgment unit to identify the target lesion.

[0038] The beneficial effects of implementing the endoscope-based digital image enhancement method and system of the present invention are as follows: the technical solution disclosed in the present invention constructs a task separation mechanism between the main control endoscope and the responsive controllable terminal, which significantly improves the recognition efficiency and intervention accuracy of suspicious lesions in complex human cavity environments, and breaks through the technical limitations of traditional single-device operation that is difficult to balance recognition and intervention. Furthermore, by extracting the coordinates of abnormal anchor points in the image and combining posture sensing information with a time synchronization mechanism, the system stability, operational safety, and remote interaction capabilities are significantly improved. Compared with traditional electronic endoscopes, it is more operable and has a significant effect on reducing diagnosis and treatment time. BRIEF DESCRIPTION OF THE DRAWINGS

[0039] Figure 1 is a flow chart of an endoscope-based digital image enhancement method of the present invention;

[0040] Figure 2 The flowchart of the spatial position information extraction process of the endoscope-based digital image enhancement method;

[0041] Figure 3 The figure is a hardware block diagram of an enhancement system for implementing an endoscope-based digital image enhancement method. DETAILED DESCRIPTION

[0042] The present invention will be further described below in conjunction with the accompanying drawings and examples so that those skilled in the art can understand and implement the present invention. However, it should be understood that these examples are intended only to illustrate the present invention and are not intended to limit the present invention. Without departing from the spirit and substance of the present invention, those skilled in the art may make various equivalent variations and substitutions to the present invention, all of which are intended to fall within the scope of protection of the present invention.

[0043] Example 1

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

[0045] Step 1: The endoscope captures n sets of lesion area images p, which are encoded to generate raw video streams, which are then transmitted to a wireless transmission unit. In this embodiment, the image resolution of the endoscope-captured images p is preferably 1920×1080 pixels, with a frame rate of 30 frames per second. After being processed by a built-in video encoding module, the images are compressed using H.265 encoding to generate n sets of independent video streams, each corresponding 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, which decodes the original video stream and transmits it to the control unit and display unit. In this embodiment, the original video stream carries frame sequence tags and timestamp information during transmission, which the medical terminal uses to time-align the image data during the decoding phase.

[0047] Furthermore, a decoding scheduling module is set up inside the medical terminal, which splits the original video stream into two data paths according to data priority: the control processing stream includes key frames, image marking information and metadata, which are decoded first and pushed to the control unit in real time; the display presentation stream contains a full-frame image sequence, which is cached in sequence 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, proceed 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 interactive interface of the display unit. The control unit initiates the identification of the abnormal anchor point and completes the release operation of the controllable terminal based on the communication instruction.

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

[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 collected by the endoscope.

[0051] In this embodiment, the characteristic marking points include the texture, color, edge and contrast abnormality features of the lesion area image p, and the comprehensive judgment basis is the abnormality scoring function S(x, y), ,in: is the image pixel coordinate; Indicates the gradient amplitude of the point, reflecting the intensity of edge change; represents the texture feature value extracted by the local binary pattern, is the texture mean of the entire 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; is the weighted coefficient of each feature, satisfying , i=1,2,...,4; when (in is the preset threshold), the point is determined to be an abnormal anchor point and its coordinate information is recorded.

[0052] In this embodiment, the two-dimensional image coordinates (x, y) of the abnormal anchor point in the endoscopic image p are extracted, and the timestamp K of the first communication instruction is read. t , through the timestamp K t Match the position information collected by the controllable terminal posture sensor at this moment, 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 groups of lesion area images q. The m groups of lesion area images q are encoded to generate enhanced video streams, which are 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 decision unit receives the enhanced video stream and extracts the x-frame enhanced image Y of the enhanced video stream. x , traverse the x frame enhanced image Y x .

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

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

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

[0059] b) Local color deviates from the central area reference value Exceeding the color difference threshold ,Right now: ;

[0060] c) The similarity between the regional grayscale contrast and texture structure and the depth model of the target lesion is higher than the contrast threshold , by the similarity function satisfy: ;

[0061] in: is the grayscale value of the pixel in image q; is the color vector in 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 will enhance the image Y i Transmitted to the display unit.

[0063] Example 2:

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

[0065] In step 5, after the endoscope identifies and marks the abnormal anchor point, it needs to extract its spatial position information and send it to the controllable terminal so that it can achieve the moving target in step 6. Figure 2 , the spatial location information extraction process includes the following steps:

[0066] Step 201: The endoscope processes the image p of the lesion area through an image analysis module, identifies image points with abnormal image features, and records their two-dimensional image coordinates in the image p, which are recorded 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 ,This timestamp serves as a reference for synchronization matching between image p and attitude sensor data.

[0068] Step 203: According to the timestamp K t, matching the spatial posture data of the endoscope at time t recorded by the controllable terminal posture sensor, and generating a homogeneous transformation matrix: ,in, Represents the rotation matrix formed by the direction angle; 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, and the pixel coordinates of the abnormal anchor point in the image p are (x, y), then the camera intrinsic parameter matrix is: ,in, are the horizontal and vertical focal lengths, is the coordinate of the principal point. Then the normalized light direction vector corresponding to the pixel point is It can be expressed as: .

[0070] Step 205: The depth value of the abnormal anchor point is recorded as , then the spatial position of the anchor point in the local coordinate system of the endoscope is: , that is, after zooming in to the true depth in the unit normalization direction, the three-dimensional point coordinates are obtained, and the reference coordinates of the abnormal anchor point in the three-dimensional space of the human cavity are It can be expressed as: ,in, It is the spatial position of the abnormal anchor point under the endoscope perspective, that is, the spatial position information of the point, which is used to set the target of the moving path.

[0071] In this embodiment, the homogeneous transformation matrix is ​​used to achieve the mapping between the image coordinate system and the three-dimensional space coordinate system. It should be understood that this technical solution achieves 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 coordinate mapping methods based on homogeneous matrices also exist 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 synergy, which distinguishes it from existing conventional processing methods.

[0072] Example 3

[0073] This embodiment details a specific method for determining target lesions, which is used to implement an endoscope-based digital image enhancement method. After the controllable terminal reaches the abnormal anchor point location in step 6, it captures an image q of the lesion area. This image q is encoded to generate an enhanced video stream, which is transmitted to the judgment unit by a wireless transmission unit. The judgment unit's task is to analyze the image features in image q and identify whether a target lesion exists based on the set judgment criteria. If so, step 9 is executed; otherwise, the process returns to step 5 to adjust the anchor point.

[0074] In step 9, the target lesion is the enhanced image Y in frame x. xThe target pathological area is captured by the controllable terminal and determined by the judgment unit to be worthy of intervention. The judgment unit performs inter-frame segmentation on image q, intercepts m frames of image, analyzes its spatial structure, color distribution, grayscale texture and other image features frame by frame, and matches them with the preset model.

[0075] In this embodiment, target lesion determination does not rely on a single image feature, but integrates three key indicators: edge mutation, color shift, and texture similarity. The determination process of the key indicators is based on logical matching of quantization thresholds and image similarity functions, supporting algorithm module deployment.

[0076] Judgment condition 1: For any pixel point (x, y) in the abnormal anchor point of the present invention, calculate the grayscale gradient amplitude of the pixel point: If there exists a region that satisfies , then the region is judged to have abnormal edge contour mutation characteristics. To set the threshold, I q It is the component channel of the image q of the lesion area of ​​the image.

[0077] Judgment condition 2: Convert image q to CIELab color space and perform color vector analysis on each pixel. Define the color vector of this point as: , with the center area color reference mean as , define color difference as: If: , then it is considered that the color shift in this area is significant and has color abnormality characteristics. 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] Judgment condition three: Suppose a candidate image area is , the depth model or template image of the target lesion is , using gray-level co-occurrence matrix to extract local texture structure With reference texture , through structural similarity function or cosine similarity function Calculate similarity: If: , then the candidate region It is highly similar to the target lesion model and has typical pathological structural characteristics, including is the comparison threshold.

[0079] In this embodiment, for judgment condition one, considering that the grayscale gradient amplitude cannot be accurately calculated under complex conditions, the following supplementary judgment condition one is used: the average gradient values ​​of the pixels corresponding to the abnormal anchor points within the 3×3 or 5×5 neighborhood range all meet the threshold condition, and the edge of the area is continuous, that is, the outline of the abnormal area must form a closed or semi-closed shape to avoid misidentification of isolated high-gradient noise points as target lesions.

[0080] Example 4

[0081] An enhancement system for implementing an endoscope-based digital image enhancement method, referring to Figure 3 , 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.

[0082] The endoscope is used to capture an image p of the lesion area, encode the image to generate a raw video stream, and release the controllable terminal based on the decision information issued by the control unit, while establishing a channel connection with the controllable terminal. In this embodiment, the endoscope is a flexible plug-in device equipped with a high-definition imaging and video encoding module. The encoding module outputs multiple video streams based on the H.265 protocol and establishes a low-latency data link with the controllable terminal using preset channel parameters.

[0083] After receiving the abnormal anchor point location information sent by the endoscope, the controllable terminal is configured to move to the abnormal anchor point location under the control of the scheduling system, capture the image q of the lesion area, and 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 dispatching system is used to perform directional control and spatial attitude adjustment on the controllable terminal based on the direction angle, depth value, and displacement vector in the position information. The dispatching system is composed of an electromagnetic coil array, and the preferred magnetic control response frequency range is 75-100Hz.

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

[0086] The judgment unit is used to enhance the image Y of frame x in the enhanced video stream. x Perform image feature analysis and determine whether a target lesion exists based on edge gradient, color shift, and texture similarity. In this embodiment, the judgment unit is integrated into the medical terminal, and the lesion area image q is processed between frames through a convolutional neural network. It should be understood that the judgment unit pre-stores a convolutional neural network model and an edge gradient threshold. , 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 control signal input capability.

[0088] 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 enhanced video stream image display that is finally transmitted back.

[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 and controllable terminal release and task scheduling of the abnormal anchor point, and coordinate the scheduling system to complete the controllable terminal path control and call the judgment unit to identify the target lesion.

[0090] The above description is only a preferred embodiment of the present invention and is not intended to limit the present invention. Any modifications, equivalent replacements and improvements made within the spirit and principles of the present invention should be included in the scope of protection of the present invention.

Claims

1. A digital image enhancement method based on endoscope, characterized in that: The following steps are involved: Step 1: The endoscope collects n sets of lesion area images p, and the n sets of lesion area images p are encoded to generate original video streams, which are 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 display unit; Step 3: If the display unit inputs the first communication instruction to the control unit, proceed to step 4; Otherwise, return to step 1; Step 4: The control unit sends a decision message 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 collects m groups of lesion area images q. The m groups of lesion area images q are encoded to generate enhanced video streams, which are 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 decision unit receives the enhanced video stream and extracts the x-frame enhanced image Y of the enhanced video stream. x , traverse the x frame enhanced image Y x ; Step 9: If there is at least one frame of enhanced image Y i If it is identified as a target lesion, go to step 10; otherwise, return to step 5, i = 1, 2, ..., x; Step 10: The wireless transmission unit will enhance the image Y i transmitting to a display unit; After the endoscope identifies and marks the abnormal anchor point, it needs to extract its spatial position information and send it to the controllable terminal so that it can achieve the moving target in step 6. The process of extracting the spatial position information includes the following steps: Step 201: The endoscope processes the image p of the lesion area through an image analysis module, identifies image points with abnormal image features, and records their two-dimensional image coordinates in the image p, which are recorded as (x, y); Step 202: After the user inputs the first communication instruction through the display unit, the control unit records the current timestamp K t ,This timestamp serves as a reference for the synchronization matching of image p and attitude sensor data; Step 203: According to the timestamp K t , matching the spatial posture data of the endoscope at time t recorded by the controllable terminal posture sensor, and generating a homogeneous transformation matrix: in, Represents the rotation matrix formed by the direction angle; represents the displacement vector of the endoscope in the body cavity; Step 204: Assume that the endoscope imaging system satisfies the pinhole camera model, and the pixel coordinates of the abnormal anchor point in the image p are (x, y), then the camera intrinsic parameter matrix is: Among them, f x 、f y is the focal length in the horizontal and vertical directions, c x 、c y As the principal point coordinate, the normalized light direction vector v1 corresponding to the pixel point can be expressed as: Step 205: The depth value of the abnormal anchor point is recorded as d, and the spatial position of the anchor point in the local coordinate system of the endoscope is: That is, the three-dimensional point coordinates are obtained after the unit normalization direction is magnified to the true depth, and the reference coordinates P of the abnormal anchor point in the three-dimensional space of the human cavity are b It can be expressed as: in, It is the spatial position of the abnormal anchor point under the endoscope perspective, that is, the spatial position information of the point, which is used to set the target of the moving path.

2. The endoscope-based digital image enhancement method according to claim 1, characterized in that: The first communication instruction is a start intervention command input by a user through the interactive interface of the display unit. The control unit starts the identification of the abnormal anchor point and completes the release operation of the controllable terminal based on the communication instruction.

3. The endoscope-based digital image enhancement method according to claim 1, characterized in that: In step 4, the decision information includes the controllable terminal release instruction and the channel parameters required for the endoscope to establish a communication channel with the controllable terminal.

4. The endoscope-based digital image enhancement method according to claim 1, characterized in that: The abnormal anchor point is a suspected pathological feature marker point identified by an image analysis module in the lesion area image p collected through an endoscope.

5. The endoscope-based digital image enhancement method according to claim 4, characterized in that: The feature markers include the texture, color, edge and contrast abnormality features of the lesion area image p, and the comprehensive judgment basis is the abnormality scoring function S(x,y). Where: (x, y) is the image pixel coordinate; Indicates the gradient amplitude of the point, reflecting the intensity of edge change; T L (x,y) represents the texture feature value extracted by the local binary pattern, is the texture mean of the entire image; C L (x,y) represents the color vector of the image in the CIELab color space, μ c is the regional color mean; D G (x,y) represents the contrast index calculated based on the gray level co-occurrence matrix, μ d is its global mean; α1, α2, α3, α4 are the weighted coefficients of each feature, satisfying ∑α i =1, i=1,2,...,4; when S(x,y)>τ, the point is determined to be an abnormal anchor point and its coordinate information is recorded, where τ is the preset threshold.

6. The endoscope-based digital image enhancement method according to claim 1, characterized in that: 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 , through the timestamp K t Match the position information collected by the controllable terminal posture sensor at that moment.

7. The endoscope-based digital image enhancement method according to claim 1, characterized in that: 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.

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

9. The endoscope-based digital image enhancement method according to claim 8, characterized in that: The target pathological area includes at least two image feature matching results: a) The edge profile mutation exceeds the set threshold τ e , that is, the gradient amplitude of the lesion edge satisfies: b) Local color deviates from the central area reference value μ c Exceeding the color difference threshold δ c , that is: ||C q (x,y)-μ c ||>δ c ; c) The similarity between the regional grayscale contrast and texture structure and the depth model of the target lesion is higher than the contrast threshold θ s , by the similarity function S(q i ,T) satisfies: S(q i ,T)>θ s ; Among them: I q (x, y) is the grayscale value of the pixel in image q; C q (x,y) is the color vector in CIELab color space; q i is the candidate region image block, T is the depth model of the target lesion; S(·) represents the similarity function.

10. An enhancement system for implementing the endoscope-based digital image enhancement method according to 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, wherein: The endoscope is used to collect images p of the lesion area, encode the images to generate original video streams, release the controllable terminal according to the decision information issued by the control unit, and establish a channel connection with the controllable terminal; 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; The scheduling system is used to perform directional control and spatial posture adjustment on the controllable terminal based on the direction angle, depth value and displacement vector in the position information; a wireless transmission unit for receiving n groups and enhanced video streams and transmitting them to the medical terminal respectively, and establishing a data communication channel between the endoscope and the controllable terminal; The judgment unit is used to enhance the image Y of frame x in the enhanced video stream. x Perform image feature analysis to determine whether a target lesion exists based on edge gradient, color shift, 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 to receive the enhanced video stream image display that is finally transmitted back; 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 coordinate the scheduling system to complete the controllable terminal path control and call the judgment unit to identify the target lesion.

Images