CN-122024304-A - Image processing method considering weak light imaging and fundus camera

CN122024304ACN 122024304 ACN122024304 ACN 122024304ACN-122024304-A

Abstract

The invention provides an image processing method considering weak light imaging and a fundus camera, which belong to the technical field of image processing, and specifically comprise the steps of determining recognition deviation conditions under different eye image characteristics based on recognition matching data, determining enhancement optimization image characteristics in the eye image characteristics based on the recognition deviation conditions, acquiring image enhancement data under different enhancement optimization image characteristics in real time, combining the recognition matching data under different enhancement optimization image characteristics, determining matching illumination intervals of the enhancement optimization image characteristics, determining similarity conditions of the matching illumination intervals of the different enhancement optimization image characteristics, combining the image enhancement data of the enhancement optimization image characteristics in the different illumination intervals, and determining an image enhancement processing strategy of the eye image characteristics, thereby realizing the determination of the matching illumination intervals under the different eye image characteristics.

Inventors

CHENG DEJI
CHENG XIANGYUN
NIU HAITAO
XU BING
Lv Xingzheng
CHENG ZIHAO

Assignees

杭州目乐医疗科技股份有限公司

Dates

Publication Date: 20260512
Application Date: 20260209

Claims (10)

1. An image processing method considering weak light imaging, which is characterized by comprising the following steps: Determining identification matching data of different disease types according to image enhancement data of a weak light image of a fundus camera, determining identification deviation conditions under different eye image characteristics according to the identification matching data when the identification matching data is determined that enhancement optimization processing is not required to be carried out on all people, and determining enhancement optimization image characteristics in the eye image characteristics according to the identification deviation conditions; Acquiring image enhancement data under different enhancement optimization image features in real time, and determining a matching light quantity interval of the enhancement optimization image features by combining identification matching data under the different enhancement optimization image features; And determining the similarity of the matched light intensity intervals of different enhancement optimization image features, and determining an image enhancement processing strategy of the eye image features by combining the image enhancement data of the enhancement optimization image features in the different light intensity intervals.
2. The image processing method considering low-light imaging according to claim 1, wherein the image enhancement data of the low-light image includes the number of users of the image enhancement of the low-light image and identification data at different disease types.
3. The image processing method considering low-light imaging according to claim 1, wherein the identification matching data of the disease type is determined according to the accuracy of the identification of the disease type of the low-light image after image enhancement.
4. The image processing method considering weak light imaging according to claim 1, wherein determining that enhancement optimization processing is not required for all persons, specifically comprises: based on the identification matching data of the disease type, determining the identification accuracy of the weak light image after image enhancement in the disease type; identifying a deviation disease type from the disease types based on the accuracy rate; and determining whether enhancement optimization processing is required to be carried out on all people according to the data of the identified deviation disease types.
5. The image processing method according to claim 4, wherein the identification deviation disease type is a disease type having an accuracy rate smaller than a preset accuracy rate threshold.
6. The image processing method considering low-light imaging according to claim 4, wherein if the number of the identified deviation disease types is greater than a preset deviation disease type number threshold, it is determined that enhancement optimization processing is required for all persons.
7. The method of image processing taking into account dim light imaging according to claim 4, wherein the ocular image feature comprises a pupil feature of the user.
8. The image processing method considering low-light imaging according to claim 1, wherein the method for determining enhanced optimized image features among the eye image features is: dividing users with the similarity of the eye image features larger than a preset similarity threshold value into the same combination based on the eye image feature data of the users; Determining the recognition accuracy of the combined user under different disease types according to the recognition deviation conditions of the combined user under different disease types; And determining whether the eye image features corresponding to the combination are enhanced optimized image features according to the identification accuracy of different disease types and the identification accuracy of the user of the combination under different disease types.
9. The image processing method for considering low-light imaging according to claim 1, wherein the method for determining the image enhancement processing strategy of the eye image features is as follows: Determining the quantity of the enhanced optimized image features belonging to the matched illumination interval in the matched illumination interval according to the similar condition of the matched illumination interval of the enhanced optimized image features, and taking the quantity of the enhanced optimized image features belonging to the matched illumination interval in the matched illumination interval as the quantity of the matched features; Determining the number of enhancement processing users in different disease types according to the image enhancement data of the enhancement optimization image features, and determining reliable observation image features in the enhancement optimization image features according to the number of disease types, of which the number of enhancement processing users is not less than a preset user number threshold; And determining an image enhancement processing strategy of the eye image features according to the number of the matching features in different matching illumination intervals and the reliable observation image features in the enhancement optimization image features.
10. A fundus camera employing an image processing method according to any one of claims 1 to 9, characterized by comprising: The image characteristic screening module is matched with the interval screening module and the enhancement processing module; Wherein the image feature screening module is responsible for determining enhanced optimized image features in the ocular image features; the matching interval screening module is responsible for determining the matching light quantity interval of the enhanced and optimized image features; the enhancement processing module is responsible for determining an image enhancement processing strategy of the eye image features.

Description

Image processing method considering weak light imaging and fundus camera Technical Field The invention belongs to the technical field of image processing, and particularly relates to an image processing method considering weak light imaging and a fundus camera. Background Fundus cameras are key devices for diagnosis and screening of ophthalmic diseases, and imaging quality of fundus cameras directly affects observation and diagnosis of key structures such as retina, optic disc, macula, blood vessels and the like by doctors. In practical clinical application, challenges such as small pupil diameter of a patient, turbid refractive medium (such as cataract), or reduced illumination light intensity required for relieving discomfort of the patient are often faced, so that effective light entering an imaging system is weak, and a 'weak light imaging' scene is formed. The prior art mainly deals with the problem of increasing illumination intensity or prolonging exposure time, but the problem may cause discomfort of patient glare and reflective shrinkage of pupils, or motion blur in dynamic imaging, which severely restricts the applicability of the device, patient compliance and image usability, so that the determination of image enhancement processing strategies under different eye image features is performed, thereby realizing the determination of the illumination interval under the eye image features applicable to all eye image features, including the dim light imaging scene, and further improving the reliability of disease identification processing becomes a technical problem to be solved urgently. Therefore, there is a need for an image processing method and fundus camera that consider low-light imaging. Disclosure of Invention In order to achieve the purpose of the invention, the invention adopts the following technical scheme: Specifically, the application provides an image processing method considering weak light imaging, which specifically comprises the following steps: s1, determining identification matching data of different disease types according to image enhancement data of a weak light image of a fundus camera, determining identification deviation situations under different eye image characteristics according to the identification matching data when the identification matching data is determined that enhancement optimization processing is not required to be carried out on all people, and determining enhancement optimization image characteristics in the eye image characteristics according to the identification deviation situations; S2, acquiring image enhancement data under different enhancement optimization image features in real time, and determining a matching illumination quantity interval of the enhancement optimization image features by combining identification matching data under the different enhancement optimization image features; S3, determining the similarity of the matched light intensity intervals of different enhancement optimization image features, and determining an image enhancement processing strategy of the eye image features by combining the image enhancement data of the enhancement optimization image features in the different light intensity intervals. The invention has the beneficial effects that: According to the method, the enhancement optimization image characteristics in the eye image characteristics are determined based on the recognition deviation condition, a progressive analysis flow of characteristic similarity grouping, intra-group disease recognition performance evaluation, multi-level performance defect analysis and feature optimization necessity judgment is followed, firstly, users are clustered according to the similarity of the eye image characteristics, secondly, the recognition accuracy of each user group on different disease types is respectively evaluated, finally, the universality, the severity and the correlation with global performance short plates of the intra-group performance defects are analyzed through a decision tree containing multiple conditions, and finally, whether the characteristic combination represented by the group belongs to the enhancement optimization image characteristics which need to be particularly optimized is judged, so that the screening of weak light imaging scenes, namely the eye image characteristics with low recognition processing reliability, is realized, data analysis processing is carried out in all illumination quantity intervals in a targeted mode, the illumination quantity interval with high recognition accuracy corresponding to the eye image characteristics can be recognized and obtained, and meanwhile, the basis is laid for recognition of the illumination quantity interval suitable for all the eye image characteristics. And determining an image enhancement processing strategy of the eye image features according to the similar conditions of the matched light intensity intervals of different enhancement optimization image