Multifunctional on-site material evidence investigation and evidence obtaining system based on intelligent robot dog

Abstract

The invention relates to a multifunctional on-site material evidence investigation and evidence obtaining system based on an intelligent robot dog, and belongs to the technical field of criminal investigation. According to the system, a robot dog is driven by a communication device and a positioner to reach a trace detection site, and collected trace detection image data are transmitted to a main control device; attitude precision adjustment is carried out through a joint encoder arranged on the angle driving framework in combination with the collected trace detection image data; calculating the actual distance between the microscope and the target trace through a trace detection distance model; based on the actual distance, the focusing driving framework is controlled to move in combination with the corresponding initial focusing parameters; and performing real-time definition evaluation on the acquired trace detection image through a definition evaluation framework, recording an optimal focusing position and related trace detection image data after the evaluation is passed, and updating a database. The multi-scene application of trace detection is realized, the problem of blurring of the trace detection image caused by environmental interference or sample movement is solved, and the focusing quality and the shooting success rate of trace detection shooting are effectively improved.

Description

TECHNICAL FIELD

[0001] The application belongs to the technical field of criminal investigation and evidence detection, and particularly relates to a multifunctional on-site material evidence investigation and evidence collection system based on an intelligent robot dog. BACKGROUND

[0002] Trace detection is of great significance to criminal investigation and other work, and can provide key clues for case cracking. Traditional trace detection relies on manual on-site investigation, and detection personnel need to carry equipment to the scene for trace collection. In the face of complex terrains such as mountainous areas, ruins or dangerous environments such as a large number of obstacles and toxic and harmful substances in the crime scene, manual investigation is low in efficiency and high in risk. The existing trace detection microscopes are mostly fixed or portable, the former can only be used in the laboratory and is difficult to quickly respond to the needs of on-site investigation, and the latter is convenient to carry but is limited in stability, focusing speed and precision in complex environments, affecting the quality of trace image collection; the existing gimbals are insufficient in stability, control precision and adaptability to trace detection requirements, and are difficult to achieve high-precision attitude adjustment and focusing control, which cannot meet the requirements of microphotography for fine observation of small objects; and the existing focusing systems are slow in focusing speed and low in precision when facing dynamic changes in the scene, and the observation equipment matched with the robot dog lacks a distance calculation and attitude compensation mechanism with multiple sensors, which leads to poor quality of collected trace detection images and affects the accuracy and reliability of trace detection results. SUMMARY

[0003] To solve the above problems in the prior art, the application provides a multifunctional on-site material evidence investigation and evidence collection system based on an intelligent robot dog, and the purpose of the application can be achieved through the following technical scheme. A multifunctional on-site material evidence investigation and evidence collection system based on an intelligent robot dog, comprising: A remote control module receives remote control instructions through a communication device integrated with the robot dog, drives the robot dog to move to a trace detection site in a complex terrain in combination with an integrated positioner, and transmits data collected based on the communication device to a host device for clarity determination; An attitude adjustment module, a microscopic gimbal integrating an angle driving architecture and a microscopic component architecture, is installed above the robot dog platform and is electrically connected to and controlled by the robot dog platform; the angle driving architecture is equipped with a joint encoder, and the attitude precision is adjusted in combination with trace image data collected by the microscopic component architecture; A laser ranging module, a laser range finder installed on the robot dog platform, acquires distance data between the robot dog and a target trace surface in real time, combines current attitude and position information of the microscopic gimbal, and calculates the actual distance between the microscope and the target trace through a trace distance model; The sharpness evaluation module selects corresponding initial focusing parameters from a pre-established focusing parameter database based on the actual distance, controls the focusing driving architecture to drive the microscope assembly architecture to move along the optical axis direction according to the initialized focusing parameters, and simultaneously starts the trace detection image sensor to collect trace detection image data in real time, and the collected trace detection image data is transmitted to the host control device for sharpness determination. The focusing driving module performs real-time sharpness evaluation on the collected trace detection image based on the sharpness evaluation architecture, calculates the sharpness index value of each trace detection image, and compares the index value with a preset sharpness threshold value; when the threshold value is exceeded, it is determined that the focusing is successful, the current position of the microscope gimbal is recorded as the best focusing position, and the focusing driving is stopped; if the sharpness index value does not reach the threshold value, the focusing driving and trace detection image collection and evaluation process is continued until the focusing is successful, and after the focusing is successful, the collected trace detection image is subjected to image quality optimization processing.

[0004] As a preferred technical solution of the present application, the angle driving architecture of the microscope gimbal in the attitude adjustment module includes three mutually perpendicular rotary joints, which respectively control the pitch angle, roll angle and yaw angle of the microscope; the rotary joints are equipped with main and auxiliary joint encoders, the main encoder provides position data under normal working condition, and the auxiliary encoder serves as a redundant backup to monitor the working state of the main encoder in real time, and when the main encoder fails, the system automatically switches to the auxiliary encoder.

[0005] Specifically, the trace detection distance model in the laser ranging module includes: The attitude angle is measured by the main joint encoder of the microscope gimbal combined with the attitude sensor, the position height difference is calculated according to the installation position of the microscope gimbal and the current state of the terrain adaptability adjustment mechanism of the robot dog platform, and the distance between the robot dog and the target trace surface is obtained by the laser range finder; Based on the optical system parameters of the microscope gimbal, corresponding parameter values are retrieved from a pre-established optical parameter database, the actual distance between the microscope and the target trace is calculated through the trace detection distance formula, and the deviation exceeds the set threshold value by comparing the average value of the current calculated actual distance with the previous calculation results, triggering the data reacquisition and calculation process.

[0006] Specifically, the focusing driving architecture in the sharpness evaluation module is controlled by the host control device of the robot dog platform, the host control device sends driving instructions to the focusing driving architecture based on the actual distance and the data in the focusing parameter database, including the rotating speed, rotating direction and rotating angle of the motor; The focus driving architecture is also equipped with a position sensor that monitors position information of the microscope component architecture along the optical axis direction in real time and feeds back the position information to the master control device, which compares the feedback position information with a target focus position to complete real-time correction of the focus process.

[0007] Specifically, the sharpness evaluation architecture in the focus driving module performs real-time sharpness evaluation on the collected trace detection image, and the steps are: The collected trace detection image is subjected to grayscale processing to convert the color trace detection image into a grayscale trace detection image, retain the brightness information of the trace detection image, remove the color information, simplify the three-dimensional color data into one-dimensional grayscale values, and generate a preprocessed trace detection image through filtering and denoising. The gradient amplitude and gradient direction of each pixel point in the preprocessed trace detection image are calculated, wherein the gradient amplitude is used to represent the intensity of the trace detection image edge, and the gradient direction is used to represent the direction information of the trace detection image edge. The gradient histogram of the trace detection image is calculated according to the gradient amplitude, the number distribution of pixel points with different gradient amplitudes is counted, and the overall sharpness of the trace detection image is evaluated by analyzing the concentration and peak position of the gradient histogram. The direction histogram of the trace detection image is constructed based on the gradient direction information, the direction distribution characteristics of the trace detection image edge are analyzed, and the sharpness index value of the trace detection image is calculated comprehensively based on the gradient amplitude and direction information.

[0008] Specifically, before calculating the sharpness index value, the trace detection image is also subjected to local sharpness evaluation, including: The trace detection image is divided into multiple local regions, the sharpness index value of each local region is calculated, and the average value of the sharpness index values of all local regions is taken as the overall sharpness index value of the trace detection image. Wherein, the division of the local region adopts an adaptive division algorithm, dynamically determines the size and position of the local region based on the principle that each local region contains sufficient edge detail information, and there is a certain overlapping region between adjacent local regions to avoid loss of edge information.

[0009] Specifically, the preset sharpness threshold in the focus driving module is classified and set according to different types of traces and microscopic observation requirements, including: An image of a trace detection sample is acquired, a training data set is constructed, images in the training data set are dynamically labeled, the dynamic labeling regionally brightens a trace detection position of the trace detection sample image, a type label code corresponding to the trace detection position is added based on a trace detection type: a preset database classification table integrates all classification types of the trace detection sample image, the type label is associated with a corresponding classification type in the database classification table, and the type label code is used as a labeling label of the trace detection sample image in the training data set. A clarity level model is generated using the labeled sample trace detection image, and trace detection image features and clarity index value distributions of different types of traces at different clarity levels are learned. Based on the generated model, the clarity threshold range corresponding to each type of trace is determined, and the middle value is taken as the preset clarity threshold, so that in the actual trace detection process, when the clarity index value of the acquired trace detection image exceeds the threshold, the trace detection image can meet the requirements of trace detection analysis.

[0010] Specifically, during the focusing process, the clarity evaluation architecture continuously acquires multiple trace detection images and calculates the clarity index values of the trace detection images, predicts the peak value position of the clarity index value corresponding to the microscopic gimbal position using a curve fitting method according to the change sequence of the clarity index value, and adjusts the moving direction of the focusing drive architecture to the peak value position direction based on the prediction result.

[0011] Specifically, after successful focusing, the acquired trace detection image is subjected to trace detection image quality optimization processing, including: Based on the clarity index value of the trace detection image and the best focusing position information of the microscopic gimbal, the blurred details in the trace detection image are restored by deconvolution processing on the trace detection image. Based on the requirements of trace detection analysis in terms of color and brightness, and in combination with the illumination parameters of the machine dog lighting device, the trace detection image is subjected to color correction and brightness adjustment; the optimized trace detection image is compared with the original acquired trace detection image, and the parameter data of the optimization processing is recorded. Based on trace detection, the host control device organizes and analyzes the acquired trace detection image data, extracts key feature information of the trace, and compares and matches the key feature information with a pre-stored trace database, to complete rapid recognition and classification of the trace.

[0012] Specifically, the host control device has a dynamic focusing compensation mechanism, including a compensation control subsystem based on posture feedback and a compensation control subsystem based on distance feedback: When the joint encoder of the angle driving architecture detects that the attitude change amount of the micro gimbal exceeds the attitude compensation threshold, the required focus compensation amount is calculated according to the direction and amplitude of the attitude change, and the focus driving architecture is controlled to drive the microscope component architecture to move a corresponding compensation distance along the optical axis direction. When the distance data change amount obtained by the laser range finder exceeds the distance compensation threshold, the actual distance between the microscope and the target trace is recalculated according to the distance change amount, and the focus driving architecture is controlled to adjust the focus based on the selected corresponding compensation focus parameter in the focus parameter database.

[0013] The beneficial effects of the present application are: The machine dog can integrate communication devices and positioners, move flexibly in complex terrain, and receive remote instructions to reach the trace detection site. The multi-joint micro gimbal can adjust the angle of the microscope in all directions, and the laser range finder and trace detection distance model can accurately calculate the distance, providing accurate basis for subsequent dynamic focusing, greatly improving the adaptability and operation efficiency of the trace detection equipment in complex sites.

[0014] The dynamic focus compensation mechanism can monitor the attitude and distance changes in real time, and compensate for the focus when the threshold is exceeded, ensuring that the focus accuracy is maintained at all times when the machine dog moves or the environment changes, and ensuring continuous and stable trace detection work.

[0015] Based on the focus parameter database, the initial focus parameter is quickly selected according to the actual distance to start focusing, the trace detection image is collected in real time by the trace detection image sensor and evaluated by the master device, the index value is calculated by the clarity evaluation architecture in multiple dimensions, and the focus is determined to be successful or not by comparing the index value with the threshold value. If it does not meet the standard, it will continue to adjust until accurate focusing is achieved, greatly improving the speed and accuracy of focusing.

[0016] The preset clarity threshold is obtained by training different types of trace samples, and the peak position is predicted by curve fitting, making the focusing process more intelligent and accurate, and ensuring that the collected trace detection image meets the trace detection analysis requirements.

[0017] The trace detection image is preprocessed by grayscale, filtering and denoising, and then the gradient amplitude and direction are calculated to construct a histogram to evaluate the clarity. The local clarity evaluation uses an adaptive division algorithm to ensure the reliability of the overall clarity index of the trace detection image, providing a high-quality trace detection image basis for trace analysis.

[0018] After successful focusing, the blurred details are restored by deconvolution using the clarity index and the best focus position information, the color is corrected and the brightness is adjusted according to the illumination parameters to optimize the trace detection image quality, and the parameters are recorded by comparing with the original trace detection image, providing more accurate and clear trace detection image data for subsequent trace detection analysis.

[0019] The rotation joint of the micro gimbal angle driving framework is equipped with a primary and secondary encoder, the primary encoder provides data normally and the secondary encoder backs up monitoring, the secondary encoder is automatically switched in case of primary fault, the trace detection task is ensured to be smooth, and the system reliability is improved.

[0020] The focusing driving framework is controlled by a master control device, receives instructions containing motor speed, direction and angle for accurate driving, is equipped with a position sensor for real-time feedback of position information, and the master control device corrects the focusing process according to the position information, so that closed-loop control is realized, and the focusing accuracy and stability are improved.

[0021] In summary, the four-dimensional cooperation of the machine dog, the gimbal, the sensor and the algorithm is realized, unmanned, high-precision and strong anti-interference operation of complex scene trace detection are realized, and revolutionary technical support is provided for the criminal investigation field. BRIEF DESCRIPTION OF DRAWINGS

[0022] In order to facilitate the understanding of those skilled in the art, the present application will be further described below with reference to the drawings.

[0023] Figure 1 A process schematic diagram of a multifunctional scene material evidence investigation and evidence collection system based on an intelligent machine dog.

[0024] Figure 2 A structure schematic diagram of an embodiment of a dynamic focusing compensation mechanism. DETAILED DESCRIPTION

[0025] In order to further illustrate the technical means and effects adopted by the present application to achieve the predetermined application purposes, the specific embodiments, structures, features and effects according to the present application are described in detail below with reference to the drawings and preferred embodiments.

[0026] Please refer to Figure 1 A multifunctional scene material evidence investigation and evidence collection system based on an intelligent machine dog, comprising: A remote control module receives remote control instructions through a communication device integrated in the machine dog, drives the machine dog to move to a trace detection site in a complex terrain in combination with a positioner integrated therein, and transmits data collected based on the communication device to the master control device for clarity determination. A posture adjustment module, a micro gimbal integrated with an angle driving framework and a microscope assembly framework, is installed above the machine dog platform and is electrically connected to and controlled by the machine dog platform; the angle driving framework is equipped with a joint encoder, and the posture accuracy is adjusted in combination with trace detection image data collected by the microscope assembly framework. A laser ranging module, a laser range finder installed on the machine dog platform, acquires distance data between the machine dog and a target trace surface in real time, combines the current posture and position information of the micro gimbal, and calculates the actual distance between the microscope and the target trace through a trace distance model. The sharpness evaluation module selects corresponding initial focusing parameters from a pre-established focusing parameter database based on the actual distance, controls the focusing driving architecture to drive the microscope assembly architecture to move along the optical axis direction according to the initialized focusing parameters, and simultaneously starts the trace detection image sensor to collect trace detection image data in real time, and the collected trace detection image data is transmitted to the host control device for sharpness determination. The focusing driving module performs real-time sharpness evaluation on the collected trace detection image based on the sharpness evaluation architecture of the host control device, calculates the sharpness index value of each trace detection image, and compares the index value with a preset sharpness threshold value; when the threshold value is exceeded, it is determined that the focusing is successful, the current position of the microscope gimbal is recorded as the best focusing position, and the focusing driving is stopped; if the sharpness index value does not reach the threshold value, the focusing driving and trace detection image collection and evaluation process is continued until the focusing is successful, and the collected trace detection image is subjected to image quality optimization processing after the focusing is successful.

[0027] Specifically, the angle driving architecture of the microscope gimbal in the posture adjustment module includes three mutually perpendicular rotary joints, respectively controlling the pitch angle, roll angle and yaw angle of the microscope; the rotary joints are equipped with main and auxiliary joint encoders, the main encoder provides position data in normal operation, and the auxiliary encoder monitors the working state of the main encoder in real time as a redundant backup, when the main encoder fails, the system automatically switches to the auxiliary encoder, ensuring the smooth progress of the trace detection task.

[0028] In this embodiment, the microscope gimbal integrating the angle driving architecture and the microscope assembly architecture is installed above the robot dog platform, and is controlled by the robot dog platform through high-speed and stable electrical connection. The angle driving architecture includes three mutually perpendicular rotary joints, respectively responsible for controlling the pitch angle (range ±45°), roll angle (range ±30°) and yaw angle (range ±60°) of the microscope, and each rotary joint is equipped with high-precision main and auxiliary joint encoders with a resolution of up to 10,000 pulses per turn. The main encoder provides accurate position data when it is working normally, and the auxiliary encoder monitors in real time, and switches immediately once the main encoder fails, ensuring the continuity of posture adjustment.

[0029] The microscope assembly architecture selects a high-resolution microscope lens with an optical magnification of 50-200 times, and is equipped with a high-sensitivity trace detection image sensor that can collect and transmit trace detection image data to the host control device in real time; the focusing driving architecture adopts a high-precision stepping motor with a minimum step angle of 0.9°, which cooperates with a position sensor to realize accurate movement control in the optical axis direction, with a movement accuracy of ±10 microns, ensuring accurate focusing.

[0030] The remote operator sends a control instruction with a trace detection site coordinate position to the robot dog through a special control terminal. After the robot dog communication device receives the instruction, it combines with the self-positioning device data to plan the optimal path, autonomously navigate in the complex terrain, avoid obstacles, and smoothly move to the trace detection site. During this period, the position, attitude and other data are transmitted back to the main control device in real time through the communication device. The main control device preliminarily determines the clarity of the transmission data and estimates the influence of the site environment on the trace detection task.

[0031] After arriving at the site, the main control device sends an attitude adjustment instruction to the microscope holder. The angle driving structure of the microscope holder starts the rotary joint according to the instruction. The main encoder feeds back the position data in real time to drive the microscope to adjust the pitch angle, roll angle and yaw angle in all directions. Combined with the trace detection image data collected by the microscope component, the attitude accuracy adjustment algorithm is used to preliminarily align the microscope to the target trace area, ensuring a good starting viewing angle for subsequent trace detection operations. The whole attitude adjustment process is completed within 2 seconds, meeting the demand for rapid response.

[0032] Specifically, the trace detection distance model in the laser ranging module includes: The joint encoder of the microscope holder measures the attitude angle in combination with the attitude sensor. The position height difference is calculated according to the installation position of the microscope holder and the current state of the terrain adaptability adjustment mechanism of the robot dog platform. The distance between the robot dog and the target trace surface is obtained by the laser range finder; Based on the optical system parameters of the microscope holder, the corresponding parameter values are retrieved from the pre-established optical parameter database. The actual distance between the microscope and the target trace is calculated by the trace detection distance formula. By comparing the actual distance calculated at present with the average value of the previous calculation results, if the deviation exceeds the set threshold, the data reacquisition and calculation process is triggered, and the working state of the related sensor is checked.

[0033] In this embodiment, the trace detection distance model satisfies the following formula: wherein, represents the straight-line distance measured by the laser range finder; represents the angle between the optical axis of the microscope holder and the horizontal plane, which is obtained by the holder inclination sensor; represents the height of the robot dog chassis from the ground; represents the extension amount of the holder lifting mechanism.

[0034] Specifically, the focus driving architecture in the clarity evaluation module is controlled by the main control device of the robot dog platform. The main control device sends driving instructions to the focus driving architecture based on the actual distance and the data in the focus parameter database, including the motor speed, rotation direction and rotation angle. The focusing driving architecture is also equipped with a position sensor which monitors the position information of the microscope assembly architecture along the optical axis direction in real time and feeds back the position information to the master control device, which compares the feedback position information with the target focusing position to complete real-time correction of the focusing process.

[0035] In the trace detection task, when the robot dog carrying the microscope holder reaches the vicinity of the target trace, the laser range finder first accurately measures the linear distance between the robot dog platform and the trace surface. At the same time, combined with the attitude data of the microscope holder and the state parameters of the terrain adaptive adjustment mechanism, the actual working distance of the microscope and the trace is accurately calculated through the trace detection distance model, with an error range of ±0.05 millimeters. According to the actual distance, the master control device quickly retrieves the matched initial focusing parameters in the focusing parameter database, including the basic speed setting of the motor as 300 revolutions per minute, the predetermined rotation direction, and the estimated rotation angle of 20 degrees, and sends the driving instructions containing these parameters to the focusing driving architecture to start the focusing process.

[0036] After receiving the instructions, the stepping motor starts to rotate according to the set speed and direction, driving the microscope assembly to move along the optical axis direction. At the same time, the trace detection image sensor of the microscope collects trace detection image data in real time at a frequency of 30 frames per second and transmits it to the master control device. In the initial stage of motor rotation, due to the rapid change of the microscope lens position, the imaging clarity of the trace in the trace detection image changes rapidly, and the clarity evaluation algorithm of the master control device analyzes the clarity index of each frame of trace detection image in real time to provide dynamic feedback for subsequent focusing correction.

[0037] Specifically, the clarity evaluation architecture in the focusing driving module performs real-time clarity evaluation on the collected trace detection image, with the following steps: The collected trace detection image is subjected to grayscale processing to convert the color trace detection image into a grayscale trace detection image, retaining the brightness information of the trace detection image and removing the color information, simplifying the three-dimensional color data into one-dimensional grayscale values, and generating a preprocessed trace detection image through filtering and denoising; The gradient amplitude and gradient direction of each pixel point in the preprocessed trace detection image are calculated, where the gradient amplitude is used to represent the intensity of the trace detection image edge, and the gradient direction is used to represent the direction information of the trace detection image edge; The gradient histogram of the trace detection image is calculated according to the gradient amplitude, the number distribution of pixel points with different gradient amplitudes is counted, and the overall clarity of the trace detection image is evaluated by analyzing the concentration and peak position of the gradient histogram; Based on the gradient direction information, the direction histogram of the trace detection image is constructed to analyze the direction distribution characteristics of the trace detection image edge, and the clarity index value of the trace detection image is calculated comprehensively based on the gradient amplitude and direction information.

[0038] In this embodiment, when the microscope is aligned with the target trace and the trace image is collected, the collected color trace image data is transmitted to the host computer in real time. First, the program performs grayscale processing on the color trace image. In this way, the three-dimensional color data is simplified to a one-dimensional grayscale trace image, while the brightness information of the trace image is retained, laying a foundation for subsequent edge detection and clarity evaluation. Then, the grayscale trace image is denoised by applying a Gaussian filter algorithm.

[0039] The gradient of the pre-processed grayscale trace image is calculated. The gradient amplitude reflects the intensity of the edge of the trace image, that is, the degree of change in pixel value at the edge; the gradient direction indicates the directional information of the edge of the trace image, which is helpful for subsequent analysis of the directional distribution characteristics of the edge of the trace image.

[0040] Based on the calculated gradient amplitude, a gradient histogram of the trace image is constructed. The range of the gradient amplitude is divided into multiple intervals, the number of pixel points in each interval is counted, and a column chart is used to display the results. By analyzing the concentration and peak position of the gradient histogram, the overall clarity of the trace image can be evaluated. If the peak value of the gradient histogram is concentrated in a higher gradient amplitude interval, it means that the edge of the trace image is clear, and the overall clarity of the trace image is high; on the contrary, if the peak value is concentrated in a lower gradient amplitude interval, the trace image may be blurred and the clarity is low.

[0041] Using the calculated gradient direction information, a direction histogram of the trace image is constructed. The range of the gradient direction is divided into multiple intervals, and the number of pixel points in each direction interval is counted. By analyzing the direction histogram, the directional distribution characteristics of the edge of the trace image can be understood. By combining the gradient amplitude and direction information, the clarity index value of the trace image is calculated by weighted average and other methods. The weight can be adjusted according to the actual application scenario and experience, for example, for trace types with important edge intensity, the weight of the gradient amplitude can be appropriately increased; for trace types with significant directional distribution characteristics, the weight of the direction information can be appropriately increased. The final clarity index value can comprehensively reflect the clarity of the trace image and provide a quantitative basis for focus adjustment.

[0042] Specifically, before calculating the clarity index value, the trace image is also subjected to local clarity evaluation, including: The trace image is divided into multiple local regions, the clarity index value of each local region is calculated, and the average value of the clarity index values of all local regions is taken as the overall clarity index value of the trace image. The local area is divided by using an adaptive division algorithm, and the size and position of each local area are dynamically determined based on the principle that each local area contains sufficient edge detail information, and there is an overlap area between adjacent local areas, so as to avoid loss of edge information.

[0043] Specifically, the preset definition threshold in the focusing driving module is classified and set according to different types of traces and microscopic observation requirements, including: An image of a trace detection sample is acquired, a training data set is constructed, and the images in the training data set are dynamically labeled. The dynamic labeling displays the trace detection position of the trace detection sample image in a region brightening manner, and adds a type label code corresponding to the trace detection position based on the trace detection type. A preset database classification table integrates all classification types of the trace detection sample image, associates the type label with the corresponding classification type in the database classification table, and takes the type label code as the labeling label of the trace detection sample image in the training data set. A definition level model is generated by using the labeled sample trace detection image, and the trace detection image features and definition index value distribution of different types of traces under different definition levels are learned. Based on the training result, the definition threshold range corresponding to each type of trace is determined, and the middle value is taken as the preset definition threshold, so that when the definition index value of the collected trace detection image exceeds the threshold in the actual trace detection process, the trace detection image can meet the requirements of trace detection analysis.

[0044] In this embodiment, during the entire focusing process, the position sensor monitors the displacement of the microscope lens along the optical axis in real time, and sends the position data to the main control device at an interval of 0.1 milliseconds. The main control device accurately compares the feedback actual position with the target focusing position, calculates the deviation therebetween, and when it is detected that the position deviation exceeds a preset threshold, the main control device dynamically adjusts the focusing driving instruction according to the size and direction of the deviation. Assuming that the current actual position of the microscope lens lags behind the target position by 3 microns, the main control device will send an acceleration instruction to the motor, increase the motor speed to 400 revolutions per minute, and appropriately adjust the rotation angle, so that the microscope lens moves quickly to the target position until the position deviation is reduced to within the threshold range, realizing real-time fine correction of the focusing process and ensuring that the focusing accuracy reaches the micron level.

[0045] Specifically, the definition evaluation architecture continuously acquires multiple trace detection images and calculates the definition index values of the trace detection images during the focusing process, predicts the peak position of the definition index value corresponding to the position of the microscopic holder based on the change sequence of the definition index values by using a curve fitting method, and adjusts the moving direction of the focusing driving architecture to the peak position direction based on the prediction result.

[0046] Specifically, after successful focusing, the collected trace detection image is subjected to trace detection image quality optimization processing, including: Based on the definition index value of the trace detection image and the best focusing position information of the microscope holder, the blurred details in the trace detection image are recovered by deconvolution processing on the trace detection image; Based on the requirements of trace analysis in color and brightness, combined with the illumination parameters of the machine dog platform lighting device, the trace detection image is subjected to color correction and brightness adjustment; the optimized trace detection image is compared with the original collected trace detection image, and the parameter data of the optimization processing is recorded; Based on trace detection, the host device organizes and analyzes the collected trace detection image data, extracts key feature information of the trace, and compares and matches with the pre-stored trace database to complete the rapid identification and classification of the trace.

[0047] Please refer to Figure 2 , specifically, the host device has a dynamic focusing compensation mechanism, including a compensation control subsystem based on attitude feedback and a compensation control subsystem based on distance feedback: When the joint encoder of the angle driving architecture detects that the attitude change of the microscope holder exceeds the attitude compensation threshold, the required focusing compensation amount is calculated according to the direction and amplitude of the attitude change, and the focusing driving architecture is controlled to drive the microscope component architecture to move a corresponding compensation distance along the optical axis direction; When the distance data obtained by the laser range finder changes by more than the distance compensation threshold, the actual distance between the microscope and the target trace is recalculated according to the distance change, and the focusing driving architecture is controlled for focusing adjustment based on the corresponding compensation focusing parameters selected from the focusing parameter database.

[0048] In this embodiment, a high-precision attitude sensor is installed on the microphotographic platform to monitor the tilt, rotation and other attitude changes of the platform in real time. At the same time, a laser range finder is provided to accurately measure the distance change between the camera and the sample. The attitude sensor and the laser range finder are connected to the host device through wireless or wired means to ensure real-time transmission of data.

[0049] When the microphotographic platform changes in attitude due to external interference, the attitude sensor detects the attitude change. If the change exceeds the preset attitude compensation threshold, the attitude feedback subsystem is triggered. According to the direction and amplitude of the attitude change, the host device calculates the required focusing compensation amount using the focusing compensation model. Then, the host device sends instructions to the focusing driving architecture to drive the microphotographic camera to move a corresponding compensation distance along the optical axis direction to offset the influence of the attitude change on the focusing position. During the movement, the position sensor feeds back the position information of the camera in real time to ensure the accuracy of the compensation movement.

[0050] Meanwhile, the laser range finder continuously monitors the distance between the camera and the sample. When the distance variation exceeds the preset distance compensation threshold, the distance feedback subsystem is activated. The host device recalculates the optimal focus position at the current distance based on the new distance data and the focus parameter curve in the focus parameter database. According to the calculation result, the host device adjusts the focus drive architecture to move the camera to the new focus position, ensuring the clarity of the trace detection image. This process effectively deals with the distance variation caused by sample position changes or platform instability through real-time data analysis and rapid focus adjustment.

[0051] After each compensation adjustment, the host device evaluates the compensation effect through trace image clarity evaluation. If the trace image clarity still does not meet the preset threshold, the host device will continue to make compensation adjustments until the trace image clarity meets the requirements. In this way, the dynamic focus compensation mechanism continuously optimizes the focus position, improving the trace detection image quality of microphotography.

[0052] The above is only the preferred embodiment of the present application, not any form of limitation on the present application, although the present application has been disclosed as above with the preferred embodiment, however, not to limit the present application, any person skilled in the art, without departing from the scope of the present application technical solution, can make some more changes or modifications of the above disclosed technical content for equivalent embodiments, but as long as it does not deviate from the technical solution content of the present application, according to the technical essence of the present application, any simplification, modification, equivalent change and modification of the above embodiments, all still belong to the scope of the present application technical solution.

Claims

1. A multifunctional on-site evidence examination and collection system based on an intelligent robot dog, characterized in that, include: The remote control module receives remote control commands through the communication device integrated in the robot dog, and drives the robot dog to move to the trace inspection site in complex terrain in combination with the integrated locator. Based on the communication device, the collected data is transmitted to the main control device for clarity determination. The attitude adjustment module, which integrates an angle driving architecture and a microscope component architecture, is mounted above the robot dog platform, electrically connected to and controlled by the robot dog platform. The angle driving architecture is equipped with a joint encoder, which is used to adjust the attitude accuracy in conjunction with the trace inspection image data acquired by the microscope component architecture. The laser ranging module acquires the distance data between the robot dog and the target trace surface in real time through a laser rangefinder installed on the robot dog platform. Combined with the current attitude and position information of the microscopic gimbal, the actual distance between the microscope and the target trace is calculated through the trace ranging model. The sharpness assessment module selects the corresponding initial focus parameters from the pre-established focus parameter database based on the actual distance, controls the focus drive architecture to drive the microscope component architecture to move along the optical axis according to the initialized focus parameters, and simultaneously starts the trace inspection image sensor to collect trace inspection image data in real time. The collected trace inspection image data is transmitted to the main control device for sharpness determination. The focus drive module performs real-time sharpness evaluation on the acquired trace inspection images based on a sharpness evaluation architecture, calculates the sharpness index value of each trace inspection image, and compares the index value with a preset sharpness threshold. If the threshold is exceeded, it is determined that the focus is successful, the current position of the microscopic gimbal is recorded as the optimal focus position, and the focus drive is stopped. If the sharpness index value does not reach the threshold, the focus drive and trace inspection image acquisition evaluation process continues until the focus is successful. After successful focus, the acquired trace inspection images are subjected to image quality optimization processing.

2. The system according to claim 1, characterized in that, The angle drive architecture of the microscopic gimbal in the attitude adjustment module includes three mutually perpendicular rotary joints, which control the pitch angle, roll angle and yaw angle of the microscope respectively. The rotary joints are equipped with two joint encoders, a main encoder and a secondary encoder. During normal operation, the main encoder provides position data, and the secondary encoder serves as a redundancy backup to monitor the working status of the main encoder in real time. When the main encoder fails, the system automatically switches to the secondary encoder to ensure the smooth progress of the trace inspection task.

3. The system according to claim 1, characterized in that, The trace detection ranging model in the laser ranging module includes: The attitude angle is measured by the main joint encoder of the microscopic gimbal combined with the attitude sensor. The position height difference is calculated according to the installation position of the microscopic gimbal and the current state of the terrain adaptability adjustment mechanism of the robot dog platform. The distance between the robot dog and the target trace surface is obtained by the laser rangefinder. Based on the optical system parameters of the microscopic gimbal, the corresponding parameter values ​​are retrieved from the pre-established optical parameter database. The actual distance between the microscope and the target trace is calculated using the trace detection distance formula. The actual distance calculated now is compared with the average value of the previous calculation results. If the deviation exceeds the set threshold, the data re-acquisition and calculation process is triggered.

4. The system according to claim 1, characterized in that, The focus drive architecture in the sharpness assessment module is controlled by the main control device of the robot dog platform. The main control device sends drive commands to the focus drive architecture based on the actual distance and data in the focus parameter database, including the motor speed, rotation direction and rotation angle. The focusing drive architecture is also equipped with a position sensor, which monitors the position information of the microscope component architecture along the optical axis in real time and feeds the position information back to the main control device. The main control device compares the feedback position information with the target focusing position and completes real-time correction of the focusing process.

5. The system according to claim 1, characterized in that, The sharpness evaluation architecture in the focus drive module performs real-time sharpness assessment on the acquired trace inspection images. The steps are as follows: The acquired trace images are processed to grayscale, converting color trace images into grayscale trace images, retaining the brightness information of the trace images, removing color information, simplifying the three-dimensional color data into one-dimensional grayscale values, and generating preprocessed trace images through filtering and noise reduction. Calculate the gradient magnitude and gradient direction of each pixel in the preprocessed trace evidence image, where the gradient magnitude is used to represent the intensity of the trace evidence image edge, and the gradient direction is used to represent the direction information of the trace evidence image edge; The gradient histogram of the trace evidence image is calculated based on the gradient magnitude, and the distribution of the number of pixels with different gradient magnitudes is statistically analyzed. The overall clarity of the trace evidence image is evaluated by analyzing the concentration and peak position of the gradient histogram. Based on gradient direction information, a direction histogram of the trace evidence image is constructed. The directional distribution characteristics of the trace evidence image edges are analyzed. Combining gradient magnitude and direction information, the sharpness index value of the trace evidence image is calculated.

6. The system according to claim 1, characterized in that, Before calculating the sharpness index value, a local sharpness assessment of the trace evidence images is also included, including: The trace evidence image is divided into multiple local regions, the sharpness index value of each local region is calculated, and the average of the sharpness index values ​​of all local regions is taken as the overall sharpness index value of the trace evidence image. The local region is divided using an adaptive partitioning algorithm. Based on the principle that each local region contains sufficient edge detail information, the size and position of the local region are dynamically determined, and there is a certain overlap between adjacent local regions to avoid the loss of edge information.

7. The system according to claim 1, characterized in that, The preset sharpness threshold in the focusing drive module is set according to different types of traces and microscopic observation requirements, including: Acquire trace evidence sample images, construct a training dataset, dynamically annotate the images in the training dataset, the dynamic annotation brightens the trace evidence location in the trace evidence sample images, and adds type label codes corresponding to the trace evidence location based on the trace evidence type: a preset database classification table integrates all classification types of the trace evidence sample images, associates the type labels with the corresponding classification types in the database classification table, and uses the type label codes as annotation labels for the trace evidence sample images in the training dataset; A sharpness level model is generated using labeled sample trace images to learn the characteristics and sharpness index value distribution of trace images of different types of traces at different sharpness levels; Based on the generated model, the sharpness threshold range corresponding to each type of trace is determined, and the median value is taken as the preset sharpness threshold to ensure that when the sharpness index value of the collected trace images exceeds the threshold during the actual trace inspection process, the trace images can meet the requirements of trace inspection analysis.

8. The system according to claim 1, characterized in that, During the focusing process, the sharpness evaluation architecture continuously acquires multiple trace images and calculates the sharpness index value of the trace images. Based on the change sequence of the sharpness index value, a curve fitting method is used to predict the position of the microscopic gimbal corresponding to the peak position of the sharpness index value. Based on the prediction result, the moving direction of the focusing drive architecture is adjusted to the direction of the peak position.

9. The system according to claim 1, characterized in that, The image quality optimization processing method in the focus driving module is as follows: Based on the sharpness index value of the trace image and the optimal focus position information of the microscopic gimbal, the blurred details in the trace image are recovered by performing deconvolution processing on the trace image. Based on the requirements of trace evidence analysis in terms of color and brightness, and combined with the illumination parameters of the lighting device of the robot dog platform, the trace evidence images are color corrected and brightness adjusted; the optimized trace evidence images are compared with the original acquired trace evidence images, and the parameter data of the optimization process are recorded. Based on trace detection, the main control device organizes and analyzes the collected trace detection image data, extracts key feature information of the traces, and compares and matches them with the pre-stored trace database to complete the rapid identification and classification of traces.

10. The system according to claim 1, characterized in that, The main control device has a dynamic focusing compensation mechanism, including a compensation control subsystem based on attitude feedback and a compensation control subsystem based on distance feedback: When the joint encoder of the angle drive architecture detects that the attitude change of the microscopic gimbal exceeds the attitude compensation threshold, it calculates the required focus compensation amount according to the direction and magnitude of the attitude change, and controls the focus drive architecture to drive the microscopic component architecture to move the corresponding compensation distance along the optical axis. When the change in distance data acquired by the laser rangefinder exceeds the distance compensation threshold, the actual distance between the microscope and the target trace is recalculated based on the change in distance. Then, based on the corresponding compensation focus parameters selected from the focus parameter database, the focus drive architecture is controlled to adjust the focus.

Images