An image quality feedback-based non-mydriatic fundus imaging automatic parameter closed-loop control method and system

Abstract

The present application belongs to the technical field of fundus imaging control and end-side preview control, and relates to a non-mydriatic fundus imaging automatic parameter closed-loop control method and system based on image quality feedback. In view of the problem that the exposure, focusing, illumination and gain parameters in the preview stage of a handheld non-mydriatic fundus camera are difficult to be set to a suitable state at one time, the method continuously collects preview frames in the preview stage, extracts a quality feedback vector including definition, reflection, exposure, coverage and jitter, and according to the types of feedback that do not meet the standards and the priority of parameters, adjusts the exposure time, gain, focusing position and illumination intensity in linkage, and limits the amplitude of single adjustment; when the comprehensive quality score and the key single minimum allowable threshold value both meet the requirements and the continuous preset frame number meets the requirements, formal shooting is triggered. The present application is suitable for preview control and formal shooting triggering of a handheld non-mydriatic fundus camera.

Description

Technical Field

[0001] This invention relates to the fields of fundus imaging control, end-to-end control of handheld non-mydriatic fundus cameras, and closed-loop control of image quality feedback, and particularly to an automatic parameter closed-loop control method and system for non-mydriatic fundus imaging based on image quality feedback. Background Technology

[0002] In handheld non-mydriatic fundus imaging, user operating distance, camera posture, changes in ambient light, and individual differences in the subject can all make it difficult to set exposure, focus, illumination, and gain parameters to an appropriate state in one go. Current practices often treat automatic exposure, automatic focus, and illumination control as independent local functions, lacking a parameter closed-loop mechanism that provides unified feedback based on image quality results.

[0003] In this type of equipment, optimizing a single parameter may also cause other quality aspects to deteriorate. For example, increasing the exposure time may aggravate the effects of handshake, increasing the gain may increase noise, and increasing the illumination intensity may exacerbate corneal reflection occlusion. Optimizing only a single parameter locally can easily result in situations where the exposure is appropriate but the sharpness is insufficient, the composition is acceptable but the reflection is too strong, or the sharpness is sufficient but the shaking is still significant.

[0004] For non-mydriatic fundus imaging, the decision to trigger a formal shot is not based on a single parameter, but rather on a comprehensive state comprised of multiple quality factors, including sharpness, reflection, exposure, coverage, and camera shake. Among these, conditions such as reflection obscuring the macula or optic disc, insufficient coverage of critical areas, and excessive sustained camera shake should be considered prohibitive conditions for triggering a formal shot. Therefore, a method and system are needed that can use preview frame image quality feedback as the core to perform joint closed-loop adjustment of exposure time, gain, focus position, and illumination intensity. Summary of the Invention

[0005] The purpose of this invention is to provide an automatic parameter closed-loop control method and system for non-mydriatic fundus imaging based on image quality feedback. By continuously acquiring preview frames during the preview stage, each preview frame is bound to the current exposure time, gain, focus position, and illumination intensity parameters. Quality feedback vectors such as sharpness, reflection, exposure, coverage, and jitter are extracted from the preview frames. Then, the exposure time, gain, focus position, and illumination intensity are adjusted in conjunction with the type of feedback that does not meet the standard and the parameter priority to form an automatic parameter closed-loop control before the actual shooting.

[0006] To achieve the above objectives, the present invention adopts the following technical solution: The system includes an image acquisition module, a quality feedback calculation module, a parameter decision module, a parameter execution module, an automatic exposure control module, an automatic focus control module, an illumination intensity control module, a gain control module, and a formal shooting trigger module. The image acquisition module is used to acquire preview frames and record current imaging parameters; the quality feedback calculation module is used to extract quality feedback vectors from the preview frames; the parameter decision module is used to generate parameter adjustment decisions based on the quality feedback vectors, formal shooting threshold groups, and key single minimum allowable thresholds; the parameter execution module is used to call the automatic exposure control module, the automatic focus control module, the illumination intensity control module, and the gain control module to perform parameter adjustments; the formal shooting trigger module is used to trigger formal shooting when a preset number of consecutive preview frames meet the formal shooting conditions.

[0007] In this invention, the quality feedback vectors can represent different imaging states. Sharpness feedback reflects the clarity of retinal details, blood vessel edges, and optic disc boundaries; reflection feedback reflects the degree of occlusion of the optic disc or macula by bright areas; exposure feedback reflects whether the overall brightness is within an appropriate range; coverage feedback reflects whether the optic disc and macula simultaneously fall within the target coverage area; and jitter feedback reflects the relative instability of the camera or the examined eye during the preview stage.

[0008] In a preferred embodiment, the system employs a dual-condition judgment triggered by both the overall quality score and the minimum allowable threshold for key individual items. If the overall quality score meets the preset trigger line, but the reflection feedback indicates that the reflection is still obscuring the central area of ​​the macula, or the coverage feedback indicates that the key structure is not within the target coverage area, the system will not perform a formal shot, but will continue to maintain the closed-loop adjustment state or prompt the user to re-align.

[0009] Compared with existing technologies, this invention has at least the following beneficial effects: 1. By forming a closed loop through multiple quality feedback vectors, parameter priority decision-making, step size limitation, and abnormal exit strategy, the consistency of formal shooting triggering is improved; 2. By jointly adjusting multiple parameters, the local optimum problem caused by single parameter optimization is reduced; 3. By iterating through continuous preview frames, it can adapt to dynamic changes in handheld non-mydriatic scenarios; 4. It can work in conjunction with multi-frame selection, readability determination, and composition guidance to form a complete acquisition control chain; 5. By prioritizing interception of key single-item thresholds and abnormal exit conditions, the stability and engineering feasibility of closed-loop control can be improved. Attached Figure Description

[0010] Figure 1 This is a schematic diagram of the overall system structure of the present invention.

[0011] Figure 2 This is a schematic diagram of the closed-loop control method of the present invention.

[0012] Figure 3 This is a schematic diagram illustrating the comparison between quality feedback and threshold in this invention.

[0013] Figure 4 This is a schematic diagram illustrating the parameter decision-making and execution process of the present invention. Detailed Implementation

[0014] Example 1: System Overall Structure like Figure 1 As shown, the system of this invention includes an image acquisition module 1, a quality feedback calculation module 2, a parameter decision module 3, a parameter execution module 4, an automatic exposure control module 5, an automatic focus control module 6, an illumination intensity control module 7, a gain control module 8, and a formal shooting trigger module 9. The image acquisition module 1 is responsible for acquiring preview frames and recording current imaging parameters; the quality feedback calculation module 2 is responsible for calculating feedback quantities such as sharpness, reflection, exposure, coverage, and shake; the parameter decision module 3 is responsible for generating parameter adjustment decisions based on the feedback quantities; the parameter execution module 4 is responsible for driving exposure, focus, illumination, and gain control; and the formal shooting trigger module 9 is responsible for triggering shooting after a preset number of consecutive preview frames meet a threshold.

[0015] Example 2: Closed-loop control process like Figure 2 As shown, the system first enters the preview imaging state and acquires a preview frame, while simultaneously recording the exposure time, gain, focus position, and illumination intensity parameters corresponding to the preview frame. Then, it extracts the sharpness feedback, reflection feedback, exposure feedback, coverage feedback, and shake feedback to form a quality feedback vector. After that, the quality feedback vector is compared with the formal shooting threshold group and the minimum allowable threshold for key individual items.

[0016] If the conditions for formal shooting are not met, the parameter decision module 3 generates parameter adjustment decisions based on the type and priority of the non-compliant feedback quantity. The parameter execution module 4 performs linked adjustments on at least two of the following: exposure time, gain, focus position, and illumination intensity. Then, a preview frame with updated parameters is collected and re-evaluated, thus forming a closed loop. When the quality feedback vector meets the formal shooting threshold group for a consecutive preset number of frames and the key feedback quantity does not trigger the prohibition condition, the formal shooting trigger module 9 triggers formal shooting.

[0017] Example 3: Quality Feedback and Threshold Comparison like Figure 3 As shown, the sharpness feedback 10, reflection feedback 11, exposure feedback 12, coverage feedback 13, and jitter feedback 14 are all input into the overall quality score 15 and then enter the threshold comparison unit 16. When the threshold comparison unit 16 determines that both the overall quality score and the minimum allowable threshold of the key individual items meet the conditions, it outputs a qualified state 17; when any key individual item feedback does not meet the minimum allowable threshold, it outputs a unqualified state 18.

[0018] The reflectivity feedback 11 may include the area ratio of the high-brightness saturation region, the location of the reflectivity region, the reflectivity intensity, and the overlap ratio between the reflectivity region and the viewing disc region or macula region. When the overlap ratio exceeds a preset reflectivity occlusion threshold, even if the overall quality score 15 reaches the preset trigger line, the threshold comparison unit 16 will still output a substandard state 18. The coverage feedback 13 can be generated based on the coverage ratio of the viewing disc region and macula region within the target coverage area. When the coverage ratio of either region is lower than the minimum coverage threshold, the formal shooting trigger is delayed.

[0019] Example 4: Parameter Decision and Execution like Figure 4 As shown, parameter decision table 19 generates different adjustment strategies based on different quality feedback. For example, when sharpness is insufficient, the focus position is adjusted first; when reflection is too strong, the illumination intensity is reduced first or the illumination sequence is changed; when exposure is underexposed, the exposure time is increased and the gain is increased after the exposure time reaches its limit; when exposure is overexposed, the exposure time, gain, or illumination intensity is reduced; when coverage is insufficient, the preview state is maintained and the formal shooting trigger is delayed. Parameter execution result illustration 20 shows that after the corresponding parameters are updated, the preview evaluation stage is entered again until the formal shooting conditions are met or the abnormal exit condition is triggered.

[0020] Example 5: Parameter Step Size and Oscillation Suppression Strategy In one implementation, different preset step sizes or scaling factors are set for exposure time, gain, focus position, and illumination intensity. For example, the focus position uses a small step size to continuously approach the sharp peak, the illumination intensity uses a graded step size to avoid abrupt changes in reflection, and the gain uses an upper limit constraint to prevent rapid noise amplification. The parameter execution module 4 limits the parameter variation range in a single iteration to avoid closed-loop oscillation caused by excessive adjustment.

[0021] Example 6: Abnormal Exit and Collaboration Strategy In one implementation, the system sets a maximum number of iterations. When the number of iterations exceeds the preset number and the formal shooting threshold is not reached, the closed-loop control is exited and the recomposition or reshoot process is initiated. In another implementation, when the target fundus area is lost, the preview frame continues to shake beyond the limit, or the key area is obscured by reflection for a long time, the system directly terminates the current closed loop and prompts the user to re-align.

[0022] The automatic parameter closed-loop control of this invention can also work in conjunction with multi-frame selection, image readability determination, and composition guidance. Specifically, it can prioritize delaying the triggering of formal shooting when the coverage conditions are not met; continue closed-loop adjustment or prompt for reshooting when the readability score fails; and use the stable parameters obtained from closed-loop control to generate a higher-quality set of candidate frames when multiple candidate images exist.

Claims

1. A closed-loop automatic parameter control method for non-mydriatic fundus imaging based on image quality feedback, characterized in that, The process includes the following steps: S101, entering preview imaging mode at the end of a handheld non-mydriatic fundus camera, acquiring preview frames, and recording the exposure time, gain, focus position, and illumination intensity parameters corresponding to the preview frame; S102, extracting a quality feedback vector based on the preview frame, wherein the quality feedback vector includes at least sharpness feedback, reflection feedback, exposure feedback, and coverage feedback, and generating a shake feedback based on continuous preview frames or inertial measurement unit data; S103, comparing the quality feedback vector with a set of thresholds for formal shooting, and determining if any key feedback value in sharpness, reflection occlusion, key area coverage, or shake has not reached the minimum threshold. S104. When the allowable threshold is reached, formal shooting is prohibited; S105. When the formal shooting conditions are not met, parameter adjustment decisions are generated based on the type and priority of the substandard quality feedback quantity; S106. According to the parameter adjustment decisions and the single change range limit, at least two of the exposure time, gain, focus position and illumination intensity are adjusted in a linked manner; S107. Preview frames after parameter updates are collected and quality feedback vectors are extracted again. The comparison, decision and adjustment steps are executed in a loop to form automatic parameter closed-loop control; S108. When a preset number of consecutive preview frames meet the formal shooting threshold group and the key feedback quantities do not trigger the prohibition conditions, formal shooting is triggered.

2. The method according to claim 1, characterized in that, The sharpness feedback is obtained based on at least one of Laplacian variance, high-frequency energy, edge intensity, or blood vessel edge sharpness; the exposure feedback is obtained based on at least one of brightness mean, histogram distribution, overexposure ratio, underexposure ratio, or dark area ratio; the jitter feedback is obtained based on at least one of the angular velocity, acceleration statistics, displacement between adjacent preview frames, or motion blur estimate output by the inertial measurement unit.

3. The method according to claim 1, characterized in that, The reflected light feedback includes the area ratio of the high-brightness saturation region, the location of the reflected region, the reflected light intensity, and the overlap ratio between the reflected region and the viewing disc region or macula region; when the overlap ratio exceeds the preset reflected light occlusion threshold, formal shooting is prohibited.

4. The method according to claim 1, characterized in that, The coverage feedback is generated based on the coverage ratio of the optic disc area and the macula area within the target coverage range; when the coverage ratio of the optic disc area or the macula area is lower than the minimum coverage threshold, the preview state is maintained and the formal shooting trigger is delayed.

5. The method according to claim 1, characterized in that, The parameter adjustment decisions include: prioritizing the adjustment of the focus position when the sharpness is insufficient; prioritizing the reduction of the illumination intensity or the change of the illumination sequence when the reflection is too strong; prioritizing the increase of the exposure time when the exposure is underexposed and increasing the gain after the exposure time reaches the upper limit; reducing at least one of the exposure time, gain, or illumination intensity when the exposure is too strong; and maintaining the preview state and delaying the triggering of the formal shooting when the coverage is insufficient.

6. The method according to claim 1, characterized in that, The parameter adjustments are performed using preset step size, preset ratio, or graded step size methods. Upper limits, lower limits, and single-iteration change limits are set for exposure time, gain, focus position, and illumination intensity to reduce closed-loop oscillation.

7. The method according to claim 1, characterized in that, The formal shooting threshold group includes a sharpness threshold, a reflection occlusion threshold, an exposure threshold, a coverage threshold, and a shake threshold. Formal shooting is triggered by the overall quality score and at least one key single-item minimum allowable threshold. If the key single-item minimum allowable threshold is not met, formal shooting will not be triggered even if the overall quality score meets the standard.

8. The method according to claim 1, characterized in that, The closed-loop control triggers formal shooting when the formal shooting conditions are met. It terminates the current closed loop when the maximum number of iterations is reached, the target fundus is detected to be lost, the preview frame is detected to be jittered beyond the limit, or the key area is detected to be blocked by reflection for a long time, and outputs prompts to re-align, reshoot, or keep the preview.

9. A closed-loop automatic parameter control system for non-mydriatic fundus imaging based on image quality feedback, characterized in that, The system includes an image acquisition module (1), a quality feedback calculation module (2), a parameter decision module (3), a parameter execution module (4), an automatic exposure control module (5), an automatic focus control module (6), an illumination intensity control module (7), a gain control module (8), and a formal shooting trigger module (9). The image acquisition module (1) is used to acquire preview frames and record current imaging parameters. The quality feedback calculation module (2) is used to extract quality feedback vectors from preview frames. The parameter decision module (3) is used to generate parameter adjustment decisions based on the quality feedback vectors, formal shooting threshold groups, and key single-item minimum allowable thresholds. The parameter execution module (4) is used to call the automatic exposure control module (5), the automatic focus control module (6), the illumination intensity control module (7), and the gain control module (8) to perform closed-loop adjustment. The formal shooting trigger module (9) is used to trigger shooting when a preset number of consecutive preview frames meet the formal shooting conditions.

10. The system according to claim 9, characterized in that, The system is deployed on the end of a handheld non-mydriatic fundus camera and is linked with at least one of a multi-frame selection module, an image readability determination module, or a composition guidance module; wherein, when the composition guidance module determines that the optic disc region or macula region is insufficiently covered, the formal shooting trigger module (9) delays the formal shooting.

Images