CN-122018136-A - Method, apparatus, storage medium and program product for assisting in microscope zoom assembly

CN122018136ACN 122018136 ACN122018136 ACN 122018136ACN-122018136-A

Abstract

The invention discloses a method, equipment, a storage medium and a program product for assisting in assembling a microscope zoom body, and belongs to the fields of microscope assembling and image processing. The invention scores the imaging effect by using the deep learning model, guides engineers to assemble the zoom body based on the score, solves the problems that the judgment standard is not uniform, nuances cannot be observed and the like caused by the current engineers to conduct adjustment by means of naked eye observation, not only can improve the accuracy of adjustment, but also can improve the efficiency of adjustment.

Inventors

YANG FEI
ZHANG SONG
TANG HAN
QIAN HAO
ZHANG YA
LEI LU
WANG HAIJIAN

Assignees

南京木木西里科技有限公司
南京凯视迈科技有限公司
南京木木智造科技有限公司

Dates

Publication Date: 20260512
Application Date: 20260318

Claims (7)

1. A method for assisting microscope zoom body assembly is characterized by constructing and training a depth learning model for evaluating imaging effect of images in an input model, in the zoom body adjustment process, a camera collects imaging images in real time and inputs the imaging images into the trained depth learning model, the depth learning model scores the imaging effect of each input image, a certain frame of image is set as S n+1 , and whether two continuous frames of images are changed or not is judged by calculating the difference value between the score S n+1 of the frame of image and the score S n of the previous frame of image, if Indicating that the two frames of images are changed, otherwise, indicating that the two frames of images are not changed, wherein If the current frame image changes, recording the current frame image as S e1 , recording S e2 , S e3 , S e4 , setting the image with the changed image as S en , comparing S en , S en-1 , S en-2 , if so The three values are the same trend, and the adjustment of the zoom body is considered to be needed to be continuously adjusted towards the current direction, if The three values are shown as different trends, and the zoom body is considered to be close to the optimal imaging effect position at the moment, and the current position needs to be changed into finer fine adjustment.
2. The method for assisting in assembling a microscope zoom body according to claim 1, wherein the deep learning model comprises a feature extraction module and a scoring estimation module which are sequentially connected, the feature extraction module performs feature extraction on an input image and transmits the feature extraction to the scoring estimation module, and the scoring estimation module adopts two fully connected layers and outputs a node as scoring output.
3. The method for assisting microscope zoom assembly according to claim 1, wherein N images with different imaging effects are acquired, the imaging effects of the images are scored, each image is randomly sent to M vision engineers for scoring the imaging effects, a highest score and a lowest score are removed, an average value of the remaining scores is taken as a final score of the image, and a set of the scored images is taken as a training set of the deep learning model.
4. The method of assembling a zoom lens of a microscope according to claim 1, wherein the trend of the image, the image score by the deep learning model, and the continuous three image scores are displayed by a display.
5. A computer device comprising a processor and a memory, the memory storing a computer program, the processor for executing the computer program to perform the method of assisting in microscope zoom assembly of any one of claims 1-4.
6. A computer storage medium storing a computer program which, when executed on a processor, performs a method of assisting in microscopic zoom assembly according to any of claims 1-4.
7. A computer program product comprising a computer program or instructions which when executed by a processor performs a method of assisting in microscope zoom assembly according to any one of claims 1 to 4.

Description

Method, apparatus, storage medium and program product for assisting in microscope zoom assembly Technical Field The invention belongs to the field of microscope assembly and image processing, and particularly relates to a microscope zoom body assembly technology. Background The zoom imaging system of a microscope is simply referred to as a zoom. The assembly of the varistors requires the assembly of the finished lenses into the corresponding finished structures. Because the machining precision of the structural member is limited and cannot completely meet the theoretical design requirement, when designing the structural member, a plurality of manually adjustable settings (such as adjusting the lens to move left and right by a knob) are usually designed to finely adjust the position of each lens in the structure. During assembly, the assembly engineer manually fine-tunes the settings and slightly moves each lens to achieve the best imaging result that the human eye is required to observe. Typically, there are tens of lenses in a zoom, and a slight movement of each lens may require corresponding fine adjustments of the other lenses, which typically takes a lot of time. The imaging effect of the zoom body is mainly influenced by factors such as spherical aberration, chromatic aberration, field curvature and the like on an optical imaging system, and after design and shaping, the influences are reduced as much as possible by an assembly and adjustment process, and the optimal imaging effect is realized. As shown in fig. 1, (a) shows an imaging effect when the adjustment is not good, it can be seen that chromatic aberration and blurring of different colors are obviously present in the peripheral area of the image, and (b) shows an imaging effect when the adjustment is good, it can be seen that the imaging effect is good. The existing assembly method mainly depends on experience of an assembly engineer, and during adjustment, imaging effects in a picture are continuously observed by naked eyes, so that manual fine adjustment is performed. The method mainly has the problems that firstly, naked eye observation is not objective, subjective evaluation of a person is not standard, subjective evaluation standards of the person at different times are not uniform, secondly, when fine adjustment is carried out after the naked eye observation, fine adjustment is difficult to detect, the manual fine adjustment does not have a good guiding direction, thirdly, on the basis of the first two reasons, a great deal of time is spent on engineering by the fine adjustment method based on the naked eye observation, the whole adjustment time is longer, the efficiency is lower, and fourthly, the finally adjusted zoom body is possibly not in an optimal state due to inaccuracy of the naked eye observation, so that the optimal imaging effect is not achieved. Disclosure of Invention In order to solve the technical problems mentioned in the background art, the invention provides a method, a device, a storage medium and a program product for assisting in assembling a microscope zoom body. In order to achieve the technical purpose, the technical scheme of the invention is as follows: In the zoom body adjustment process, a camera collects imaging images in real time and inputs the imaging images into the trained deep learning model, the deep learning model scores the imaging effect of each input image, a certain frame of image is set as S n+1, and whether two continuous frames of images are changed or not is judged by calculating the difference value between the score S n+1 of the frame of image and the score S n of the previous frame of image, if yes, the method for assisting the zoom body assembly of a microscope comprises the steps of Indicating that the two frames of images are changed, otherwise, indicating that the two frames of images are not changed, whereinIf the current frame image changes, recording the current frame image as S e1, recording S e2 , Se3 , Se4 , setting the image with the changed image as S en, comparing S en , Sen-1 , Sen-2, if soThe three values are the same trend, and the adjustment of the zoom body is considered to be needed to be continuously adjusted towards the current direction, ifThe three values are shown as different trends, and the zoom body is considered to be close to the optimal imaging effect position at the moment, and the current position needs to be changed into finer fine adjustment. Further, the deep learning model comprises a feature extraction module and a scoring estimation module which are sequentially connected, the feature extraction module performs feature extraction on an input image and transmits the feature extraction to the scoring estimation module, and the scoring estimation module adopts two full-connection layers and outputs a node as scoring output. Further, N images with different imaging effects are acquired, the imaging effects of the images are scored, each image is randomly se