Optic Disc Localization and Focusing Method Applicable to Fundus Cameras

Abstract

Optic nerve head positioning and focusing method applicable to fundus cameras, which relates to the field of digital image processing technology. It solves the problem that the existing fundus autofocus system can only determine the region of interest based on the already captured fundus images. Due to the inevitable movement of the human eye, the positioning and focusing processes of the region of interest cannot be carried out simultaneously, resulting in inaccurate focusing. In the present invention, while the camera continuously adjusts the focusing distance, it also takes pictures of the fundus of the human eye, obtains a series of fundus images with different focusing distances and estimates the position of the region of interest, and calculates the focusing evaluation function value in the local area around it; by predicting the position of the optic nerve head and estimating the position of the corresponding region to be detected, and calculating its focusing function value. Repeat multiple times until the focusing evaluation function reaches the maximum value, and determine the corresponding focusing distance. This can ensure that the positioning and focusing of the region of interest can be carried out simultaneously, ensure accurate focusing, and obtain clearer images.

Description

Technical Field

[0001] The invention relates to the technical field of digital image processing, and in particular to an optic disc positioning and focusing method suitable for a fundus camera. Background Art

[0002] In the information age, many emerging technologies have brought many conveniences to people. At the same time, people often overuse their eyes, making fundus health problems more and more prominent. The detection rates of diseases such as high myopia, glaucoma, and age-related macular degeneration have increased. In addition, other chronic diseases (hypertension, diabetes, etc.) can also be reflected in the condition of fundus blood vessels. Therefore, fundus cameras have become the main equipment for hospitals at all levels to check human eye health and other related diseases. It is usually used to obtain images of the patient's retina for ophthalmologists to interpret and then determine the type of patient. However, before interpretation, in order to obtain a clear fundus image, first manually align the pupil, adjust the object distance, select the focus area (i.e., the area of ​​interest), manually focus, and finally obtain the fundus image. In the actual examination process, the patient's eyes will move, so doctors usually spend a long time repeating the above operations until a clearer image is obtained.

[0003] The existing fundus autofocus system can only determine the region of interest based on the captured fundus image. However, due to the inevitable movement of the human eye, the positioning and focusing processes of the region of interest cannot be performed simultaneously, resulting in inaccurate focusing. Summary of the invention

[0004] The present invention aims to solve the problem that the existing fundus automatic focusing system can only determine the region of interest based on the fundus image that has been taken, and the positioning and focusing processes of the region of interest cannot be carried out simultaneously due to the inevitable movement of the human eye, resulting in inaccurate focusing. An optic disc positioning and focusing method suitable for a fundus camera is provided to solve the problem that the existing fundus automatic focusing system can only determine the region of interest based on the fundus image that has been taken, and due to the inevitable movement of the human eye, the positioning and focusing processes of the region of interest cannot be carried out simultaneously, resulting in inaccurate focusing.

[0005] The optic disc positioning and focusing method applicable to the fundus camera is implemented by the following steps:

[0006] Step 1, acquiring multiple fundus images, using an image analysis unit to process the multiple fundus images, obtaining the position of the optic disc and calculating the focus evaluation function value of the corresponding region of interest;

[0007] Step 2: using the optic disc position sequence as input, using a prediction unit to predict the position of the optic disc in the next frame image, and determining the position of the region of interest;

[0008] Step 3, selecting a local area from the position of the region of interest determined in step 2, and calculating a focus evaluation function value of the local area;

[0009] Step 4: The fundus camera acquires the next frame of image, processes it to obtain the position of the optic disc in this frame, updates it through the prediction unit, determines the corresponding region of interest, and then calculates the focus evaluation function value.

[0010] Step 5: Repeat Step 4 until the focus distance corresponding to the maximum value of the focus function evaluation function is reached, and the automatic focusing is completed.

[0011] Advantages of the present invention:

[0012] Existing fundus cameras need to irradiate with pulsed strong light to obtain fundus images, which can easily cause discomfort to patients, and can only obtain single fundus images and cannot display conditions such as microvascular blood flow. Therefore, it is difficult to reflect the development process of patients' chronic diseases. The present invention helps to realize real-time acquisition of clear fundus images under the condition of non-mydriasis, can form a video stream, can dynamically reflect conditions such as microvascular blood flow, and is more helpful for understanding the development process of patients' chronic diseases.

[0013] In the method of the present invention, while the camera continuously adjusts the focus distance, it also takes pictures of the fundus of the human eye. After obtaining a series of fundus images at different focus distances, it analyzes and processes them to obtain the position of the optic disc in each image, and then estimates the position of the region of interest, and calculates the focus evaluation function value in the local area around it. Finally, based on the obtained sequence of optic disc positions, it predicts the position of the optic disc in the next frame of image, estimates the position of the corresponding region to be detected, and calculates its focus function value. At the same time, the system continues to acquire the next frame of image, analyzes and calculates the true position of the optic disc, and then transmits it to the prediction system for update to improve the accuracy of its prediction. Repeat this process until the focus evaluation function reaches the maximum value and the corresponding focus distance is determined. This can ensure that the positioning of the region of interest and the focusing can be carried out simultaneously, ensure accurate focusing, and obtain clearer images.

[0014] In the focusing system of the present invention, the order of the sub-units in the image analysis unit cannot be changed. Changing the order will result in abnormal analysis and processing of the image.

[0015] The prediction unit in the present invention includes predicting the position of the optic disc in the next frame of image based on the positions of the optic discs in the previous few frames of images, which is the key to ensuring that the positioning of the region of interest and the focusing can be carried out almost simultaneously.

[0016] The automatic focusing system in the present invention includes determining the focus evaluation function curve and determining the focus distance corresponding to its maximum value. Description of the Drawings

[0017] Figure 1 It is a schematic block diagram of the focusing system in the method for optic disc positioning and focusing applicable to a fundus camera according to the present invention.

[0018] Figure 2 Flow chart of the image analysis unit;

[0019] Figure 3 Effect diagram of the original fundus image;

[0020] Figure 4 For Figure 3 Effect diagram of removing overexposed area (specular reflection);

[0021] Figure 5 For Figure 4 Effect diagram of performing binary image;

[0022] Figure 6 For Figure 5 Effect diagram after performing opening operation to remove unnecessary areas;

[0023] Figure 7 For Figure 6 Effect diagram of performing hole filling;

[0024] Figure 8 For Figure 7 Effect diagram after performing roundness calculation;

[0025] Figure 9 Schematic diagram of the focusing function. Specific implementation mode

[0026] Combined with Figures 1 to 9 Describe this implementation mode, which is applicable to the optic disc positioning and focusing method of a fundus camera. The focusing method is realized through a focusing system. In order to enable the fundus camera to automatically determine the position of the region of interest in the fundus, this implementation mode provides an optic disc positioning and automatic focusing system. The optic disc is one of the most prominent features in the fundus image. Since the human eye structure is relatively fixed, once the position of the optic disc is determined, combined with the orientation information of the eye (left eye or right eye), other regions of interest (such as the macula, etc.) are also determined. In addition, when the camera acquires fundus images, the human eye often moves, which is inevitable and will lead to inaccurate focusing. Therefore, while the camera acquires a series of images, it should be able to predict the position of the optic disc in the next frame by combining the positions of the optic discs in the previous few frames, and adjust the focusing distance while predicting.

[0027] As Figure 1 shown, the focusing system includes a fundus camera, an image analysis unit for determining the position of the optic disc, a prediction unit (predicting the change trend of the pupil position), and an automatic focusing system.

[0028] As Figure 2As shown, the image analysis unit includes a saturated pixel processing unit, an adaptive binary threshold unit, an opening operation unit, a filling unit, a region attribute calculation unit, and a judgment unit.

[0029] The fundus camera is used to acquire a fundus image and determine the position of the optic disc through the image analysis unit;

[0030] The prediction unit predicts the position of the optic disc in the next frame image and the position of the region of interest according to the position of the optic disc determined by the image analysis unit;

[0031] The automatic focusing system calculates a focusing evaluation function according to the region determined by the prediction unit and feeds it back to the fundus camera for corresponding adjustment.

[0032] The image analysis unit finally determines the position of the optic disc through the saturated pixel processing unit, the adaptive binary threshold unit, the opening operation unit, the filling unit, the region attribute calculation unit, and the judgment unit in sequence;

[0033] The saturated element processing unit removes the overexposed area from the original image; then the image is binarized by the adaptive binarization unit to obtain a binary image, and at this time the binary image contains many regions of different sizes

[0034] The opening operation unit is used to retain regions within a certain size range, and then the filling unit is used to fill the holes in the region;

[0035] The region attribute calculation unit calculates the area, perimeter, and roundness of each region, and the judgment unit judges the region with the largest roundness and regards it as the optic disc.

[0036] The focusing method described in this embodiment is specifically implemented by the following steps:

[0037] 1. First, the fundus camera continuously acquires a series of (n) fundus images f1(x, y), f2(x, y)…f n (x, y) during the focusing process, and their corresponding focusing distances are l1, l2…, l n , and the image analysis unit processes and analyzes this series of fundus images. The analysis process of each fundus image is as follows:

[0038] a) First, the saturated pixel processing unit sets to zero the overexposed area (brightness exceeding 80% of the full brightness) in the original image as shown Figure 3 to avoid the interference of corneal reflection on the subsequent optic disc positioning, as shown Figure 4 ;

[0039] b) The image is binarized using an adaptive binarization threshold unit. The brighter regions are set to 1 and the darker regions are set to 0 to obtain a binary image. At this time, there are many white regions of various sizes and shapes. Some regions represent information such as the blood vessels and optic nerve disc of the fundus, and some are interference regions (such as circular frames, noise, etc.) as Figure 5 shown;

[0040] c) Since the size variation range of the optic nerve disc is relatively stable, two parameters are set, and an opening operation unit is used to perform an opening operation to remove regions that are too large or too small, as Figure 6 shown.

[0041] d) Some of the remaining regions will have "holes" that interfere with subsequent roundness calculations, so a filling unit is needed to fill the holes. As Figure 7 shown.

[0042] e) A region attribute calculation unit is used to calculate the attributes of each region: perimeter P, area S, and roundness C = 4πS / P 2 , as Figure 8 shown.

[0043] f) A judgment unit is used to regard the region with the largest roundness as the optic nerve disc and calculate its centroid, which is the position of the optic nerve disc.

[0044] II. Calculate the positions of the optic nerve discs for n fundus images according to the above process to obtain a position sequence (x1, y1), (x2, y2), … (x n , y n ), and at the same time calculate the defocus evaluation function values v1, v2, …, v n of the corresponding regions of interest. Using the position sequence as the input, let the prediction unit predict the position of the optic nerve disc in the next frame of the image, that is to determine the position of the region of interest accordingly.

[0045] III. Select a local region according to the position of the region of interest determined above and calculate its defocus evaluation function value v n+1 .

[0046] IV. At the same time, the camera is also acquiring the next frame of the image f n+1 (x, y), and performing the same analysis and processing to obtain the position (x n+1 , y n+1 ) of the optic nerve disc in this frame, updating the prediction unit to improve its prediction accuracy. The corresponding region of interest is also determined and its defocus evaluation function value v n+1 is calculated.

[0047] V. Repeat step IV until the defocus function evaluation function can determine the defocus distance corresponding to the maximum value, asFigure 9 as shown; then the process of automatic focusing is completed.

[0048] The technical features of the above-described embodiments can be combined arbitrarily. For the sake of brevity of description, not all possible combinations of the technical features in the above-described embodiments are described. However, as long as there is no contradiction in the combination of these technical features, it should be considered to be within the scope described in this specification.

[0049] The above-described embodiments merely represent several implementation manners of the present invention, and the description thereof is relatively specific and detailed. However, it should not be construed as a limitation on the scope of the invention patent. It should be noted that for those of ordinary skill in the art, without departing from the concept of the present invention, several modifications and improvements can still be made, and these all belong to the protection scope of the present invention. Therefore, the protection scope of the present invention patent shall be subject to the appended claims.

Claims

1. An optic disc positioning and focusing method applicable to a fundus camera, characterized in that: The focusing method is implemented by the following steps: Step 1: Obtain multiple fundus images, and use an image analysis unit to process the multiple fundus images to obtain the position of the optic disc and calculate the focusing evaluation function value of the corresponding region of interest; Step 2: Use the prediction unit to predict the position of the optic disc in the next frame of image with the optic disc position sequence as the input, and determine the position of the region of interest; Step 3: Select a local region from the position of the region of interest determined in Step 2, and calculate the focusing evaluation function value of the local region; Step 4: The fundus camera acquires the next frame of image, processes it to obtain the position of the optic disc in this frame, updates it through the prediction unit, determines the corresponding region of interest, and then calculates the focusing evaluation function value; Step 5: Repeat Step 4 until the focusing distance corresponding to the maximum value of the focusing evaluation function is reached, and the automatic focusing is completed.

2. The method for optic disc positioning and focusing applicable to a fundus camera according to claim 1, wherein: The specific process of obtaining the position of the optic disc in Step 1 is as follows: Step 1-1: Set the overexposed region in the fundus image to zero, and perform binarization processing to obtain a binary image; Step 1-2: Perform opening operation processing on the binary image in Step 1-1, and fill the holes in the remaining regions; Step 1-3: Calculate the attributes of each region, take the region with the largest roundness as the optic disc, and calculate its centroid, and this centroid is the position of the optic disc.

3. The method for optic disc positioning and focusing applicable to a fundus camera according to claim 1, characterized in that: The focusing method is implemented through a focusing system; the focusing system includes a fundus camera, an image analysis unit for determining the position of the optic disc, a prediction unit, and an automatic focusing system; The fundus camera is used to acquire fundus images and determine the position of the optic disc through the image analysis unit; The prediction unit predicts the position of the optic disc in the next frame of image and the position of the region of interest according to the position of the optic disc determined by the image analysis unit; The automatic focusing system calculates the focusing evaluation function according to the region determined by the prediction unit and feeds it back to the fundus camera for corresponding adjustment.

4. The method for optic disc positioning and focusing applicable to fundus cameras according to claim 3, characterized in that: The image analysis unit includes a saturated pixel processing unit, an adaptive binarization threshold unit, an opening operation unit, a filling unit, a region attribute calculation unit, and a judgment unit; The image analysis unit finally determines the position of the optic disc through the saturated pixel processing unit, the adaptive binarization threshold unit, the opening operation unit, the filling unit, the region attribute calculation unit, and the judgment unit in sequence; The saturated pixel processing unit removes the overexposed region from the original image; then the image is binarized through the adaptive binarization threshold unit to obtain a binary image. At this time, the binary image contains many regions of different sizes. The opening operation unit retains the regions within a certain size range, and then uses the filling unit to fill the holes in the regions; The region attribute calculation unit calculates the area, perimeter, and roundness of each region, and the judgment unit judges the region with the largest roundness and regards it as the optic disc.

Images