Research On Image Capture Equipment And Processing Algorithm For Facing To Retinoscopy Optometry

Eye diseases are a common problem faced by the whole world.With the development of the electronic information society,the number of patients suffering from eye diseases is increasing.Among them,the incidence of eye refractive errors such as myopia and hyperopia is extremely high.The rapid and accurate identification and suppression of refractive errors is the focus of current research.The emergence of computer optometry makes the identification of refractive errors more convenient.However,due to its complicated mechanical structure and high requirements for the patient's cooperation,the diagnostic results are not stable.As an objective optometry method,the optometry method has relatively high reliability,accuracy,and high clinical reliability.However,due to the complexity of mastering the theory of retinoscopy optometry,the long training cycle required for optometrists to fully master,and the process of optometry relying on human judgment,its promotion is limited.In this paper,based on the principle and characteristics of retinoscopy optometry,intelligent optometry equipment has been developed.Human subjective factors are eliminated,and automated optometry is realized.The diagnosis is fast and accurate,the operation is simple,easy to learn.Conducive to its further promotion.In this paper,we take the light observed by optometry as the starting point and complete the following tasks:(1)According to the optometry environment of the optometry method and the characteristics of the observed light,and according to the distribution characteristics of the gray histogram of the image,MATLAB adjustment functions are used to implement the gray transformation of the different captured images under the adaptive parameters,using Sobel calculation.The sub-and automatic threshold segmentation method segmented the mapping light from the image and used the coordinate fitting method to parameterize the mapping light feature.(2)According to the characteristics of optometry and optometry,design a suitable mechanical structure,select the control system,and select the sensor for the image acquisition module to complete the design of the image acquisition system for automatic optometry.(3)This paper proposes a quality detection algorithm suitable for retinoscopy images.It uses objective evaluation methods to classify the evaluation indicators into three categories: sharpness,pupil position,and closed-eye detection.In the evaluation of sharpness,the energy gradient method is used to calculate the gradient of the pixels of the image and obtain its sharpness value.In the pupil position degree detection,according to the position characteristics between the light and the pupil,a piece of edge-based consistent information is proposed.The evaluation method of pupil position degree uses the coincident edge of the reflected light and the pupil to fit the pupil position and make a judgment;in closed-eye detection,a multiple Hough fitting detection algorithm based on Bayesian—Monte Carlo principle is proposed,according to the Bayesian principle,the Huff circle recognition is promoted and improved,and Monte Carlo is used.The Carlow method performs random point calculations and obtains the test results.Eliminate images that do not meet any of the above three indicators,and retain high-quality images for subsequent analysis operations to ensure detection quality.(4)A mathematical model and its image processing algorithm for the relative feature parameters of the imaging method are proposed.The four observational characteristics of the reflected light,which are observed by the application of the shadow inspection method,are parametrically processed: the shadowing speed,the width of the light,the brightness of the light,and the shape of the light.Among them,in terms of shadowing speed,a mathematical model of the direct speed of shadowing generated by the fixed speed inspection mirror is proposed;in terms of the light width,corresponding calculation methods are proposed according to different light reflection forms;in terms of light brightness,three channels of R,G,and B are used.The gray value calculates the average brightness of the area and uses it as the brightness value of the light to detect the neutral state.In terms of the shape of the light,the circularity is introduced as its evaluation index to distinguish the cone crystal from the normal crystal.The simulated eye was used to map the shadowing speed and the width of the light to the diopter and perform regression processing.Use the entropy method to calculate the weight ratio of the shadowing speed and the width of the light,and the regression equation was projected to the human eye to complete its clinical application.(5)By comparing the diagnostic results obtained in this article with the standard diopter values and computer optometry results,it is found that the research method proposed in this paper and the designed mechanical device are effective,which can be performed under the premise of less time consuming and guaranteed accuracy,Realizing the inspection automation,has high practicality.