Image quality improving method based on optical fiber image transmitting bundle system

Abstract

The invention discloses an image quality improving method based on an optical fiber image transmitting bundle system. The image quality improving method comprises the four steps of adding a mask, removing honeycomb artifacts, removing gluing artifacts and removing broken bundle artifacts. In the step of mask addition, a honeycomb artifact, a broken beam artifact and a glue artifact are distinguished by adding glue and broken beam masks to the image. In the step of removing the honeycomb artifacts, the positions of the artifacts are determined in the binary image based on the mask, and the honeycomb artifacts in the image are removed in the original image based on the mask by using an edge equal difference method; in the step of removing the glue-connected artifacts, the glue-connected artifacts in the image are removed in a glue-connected mask by using a bidirectional traversal maximum value filtering method; in the step of removing the broken beam artifacts, the broken beam artifacts in the image are removed in the broken beam mask by using a bilateral equal difference assignment method. According to the method, the noise generated by the inherent structure of the optical fiber bundle in the image of the optical fiber image transmission bundle system is effectively removed.

Description

Technical Field

[0001] The invention belongs to the field of digital image processing, and in particular relates to a technology for processing noise caused by an optical fiber bundle in an image of an optical fiber image transmission bundle system. Background Art

[0002] Fiber optic image bundle is a passive optical component that can transmit images and is resistant to high temperatures, electromagnetic interference, and can be bent at multiple angles. The image acquisition system composed of fiber optic image bundles can separate the acquisition end and the imaging end to achieve passive acquisition. Based on this feature, the image acquisition system based on fiber optic image bundles can be used in some complex environments, such as medical endoscopes, complex electromagnetic environments, power systems, and small space image transmission. In addition, in the face of some extreme environments, such as strong electromagnetic environments or environments containing explosive gases, ordinary industrial cameras cannot operate normally and safely. However, the imaging system composed of fiber optic image bundles can place the passive acquisition end in the operating area and the active imaging end in the safe area, thereby ensuring its normal and safe operation.

[0003] The optical fiber image bundle is composed of multiple single-filament optical fibers, which are composed of a core with high light transmittance and a cladding with low light transmittance. This physical structure prevents light from completely passing through the optical fiber bundle to enter the back-end camera for imaging, resulting in the image captured by the camera showing neatly staggered circular light spots. The dark parts between the light spots form artifacts with an outer edge that is approximately hexagonal and an inner edge that is approximately circular, resembling a honeycomb pattern, which is a honeycomb artifact. This honeycomb artifact will reduce the overall quality of the image and seriously affect the subsequent use of the captured image. In addition, the optical fiber bundle will inevitably be damaged. On the one hand, some of the cores may break, resulting in a broken bundle, so that light cannot pass through the broken fiber, which is represented in the form of black spots on the image; on the other hand, the cladding of two adjacent rows of cores will be glued together. The two rows of cores are no longer staggered, but neatly arranged, and the artifacts between them are deeper, which results in the captured image having more obvious thin strips of dark artifacts. The method for improving the image quality of the optical fiber image bundle system aims to eliminate the image noise caused by honeycomb artifacts, broken bundles and glue by analyzing the characteristics of optical fiber images. Summary of the invention

[0004] In order to solve the above problems existing in the image of the optical fiber image transmission bundle system, a method for improving the image quality of the optical fiber image transmission bundle system includes the following steps:

[0005] S1. Adding a beam break and glue joint mask to the image: processing the acquired image and adding a mask to the image artifact part formed by the optical fiber beam break and glue joint;

[0006] S2, remove honeycomb artifacts: determine the location of honeycomb artifacts based on the image with the mask added, and use the edge arithmetic to remove the honeycomb artifacts;

[0007] S3, remove glue artifacts: use the bidirectional traversal method and the maximum filtering method to remove glue artifacts;

[0008] S4. Remove broken beam artifacts: Use the bilateral arithmetic assignment method to remove broken beam artifacts.

[0009] Optionally, the step S1 of adding a mask comprises the following steps:

[0010] S11: using the optical fiber image bundle system to collect a pure white background image P1;

[0011] S12: Set thresholds V1 and V2 (V2 <V1)、C;

[0012] S13: Binarize the pure white background image P1 based on the threshold V1, and output a binary image P2 with the same size as the original image;

[0013] S14: Calculate and output the number of 1 elements in the template;

[0014] S15: convolve the binary image P2 using the template to obtain a convolution sum;

[0015] S16: Use the convolution sum obtained in S14 to divide by the number of outputs in S15 to obtain an average value a;

[0016] S17: adding a beam-break mask to the area where a is less than a threshold C;

[0017] S18: traverse the original image with the beam break mask added;

[0018] S19: Add a glue mask to the pixel area smaller than V2 in the pure white background image P1 (outside the beam break mask area).

[0019] Optionally, the step S2 of determining the position of the honeycomb artifact based on the image with the added mask comprises the following steps:

[0020] S21: Calculate and output the number of 1 elements in the template that has the same pixel width as the core portion;

[0021] S22: using the template to convolve the binary image P2 output in S13 (skipping the mask area) to obtain the convolution sum;

[0022] S23: Use the convolution sum obtained in S22 and divide it by the number of outputs in S23 to obtain the average value t;

[0023] S24: setting a threshold T;

[0024] S25: If t < T, then determine that the edge of this region is a honeycomb artifact, and use the edge equal difference method to remove the honeycomb artifact at the same position in the original image;

[0025] S26: Calculate and output the number of elements with a value of 1 in the template that is one circle larger than that in S21;

[0026] S27: Process the image in the same way as S22 - S25 repeatedly.

[0027] Optionally, the method for removing honeycomb artifacts by the edge equal difference method in step S2 includes the following steps:

[0028] a: Determine the equal difference processing direction; when there are only two elements with a value of 1 in the same row or column, then determine its direction as horizontal or vertical, and the specific orientation is the direction where the adjacent cross-region is 0; when there are two 0 elements in the cross-neighborhood, uniformly inward; when there are three 0 elements in the neighborhood, uniformly take the direction closer to the center of the template as the equal difference processing direction; when there are no 0 elements in the neighborhood, uniformly take the direction of the line connecting to the center of the template inward as the equal difference processing direction;

[0029] b: Calculate the difference d1 between the pixels in the original image corresponding to the two elements with a value of 0 in the template along the processing direction;

[0030] c: Use the pixel adjacent to the pixel with a value of 0 in the template corresponding to the point to be processed plus the pixel difference d1, and assign it to the point to be processed.

[0031] Optionally, the method for removing glue connection artifacts in step S3 includes the following steps:

[0032] S31: Calculate the difference Δx between the maximum abscissa and the difference between the maximum ordinates in the mask region respectively;

[0033] S32: If Δx ≥ Δy, then traverse vertically;

[0034] S33: Set the center point for each column;

[0035] S34: Traverse from the leftmost and rightmost of this column towards the center point at the same time, stop traversing when reaching the center point, and then proceed to the next column after completion;

[0036] S35: If Δx < Δy, then traverse horizontally;

[0037] S36: Set the center point for each row;

[0038] S37: Traverse from the leftmost and rightmost of this row towards the center point at the same time, stop traversing when reaching the center point, and then proceed to the next row after completion;

[0039] S38: Search for the point with the largest pixel value in the 3*3 neighborhood and assign its value to the point to be processed.

[0040] Optionally, removing the broken beam artifact in step S4 comprises the following steps:

[0041] S41: taking a blank area with a width of 4 pixels on both sides of the broken beam mask, and taking the blank area as a processing area together with the broken beam mask area;

[0042] S42: Calculate the difference between the leftmost pixel value and the left and right pixel values ​​of each row in this area;

[0043] S43: Divide the difference by 17 to get the tolerance d2;

[0044] S44: Perform arithmetical assignment of values ​​to pixels in the same row.

[0045] In general, compared with the prior art, the above technical solution conceived by the present invention can achieve the following beneficial effects.

[0046] (1) The mask adding module in the method for improving the image quality of the optical fiber image transmission system proposed in the present invention can process the honeycomb artifacts, glue connection and broken bundles that have always existed in the image of the optical fiber image transmission system at one time, while ensuring that the various processing processes do not interfere with each other during the processing process, thereby avoiding multiple processing of the same area resulting in an overly blurred image;

[0047] (2) The honeycomb artifact removal module in the method for improving the image quality of the optical fiber image bundle system proposed by the present invention can achieve good processing effects even in the case where the artifact areas of different sizes are caused by irregular deformation due to compression of the fiber core by sequentially processing from small templates to large templates;

[0048] (3) The debonding module in the method for improving the image quality of the optical fiber image transmission bundle system proposed by the present invention first determines the direction of bidirectional traversal, so that during the debonding process, the points to be processed are assigned to points of nearby pixels as much as possible, thereby avoiding the situation where a maximum value affects the global bonding;

[0049] (4) The present invention proposes a broken beam removal module in a method for improving the image quality of an optical fiber image transmission bundle system. The module avoids the influence caused by the artifact area by using the bilateral arithmetic difference method. Considering the large pixel difference on both sides of the broken beam area, the pixels are assigned the arithmetic difference values ​​of the pixels on both sides to make the edge transition smooth and natural. BRIEF DESCRIPTION OF THE DRAWINGS

[0050] Figure 1 A system block diagram of the method for improving image quality based on the optical fiber image transmission bundle system provided by the present invention;

[0051] Figure 2 A strategy diagram for adding a mask in the method for improving the image quality of the optical fiber image transmission bundle system provided by the present invention;

[0052] Figure 3 Template required for adding beam-breaking mask provided by the present invention

[0053] Figure 4 A strategy diagram for removing honeycomb artifacts in the method for improving image quality based on the optical fiber image bundle system provided by the present invention;

[0054] Figure 5 The template required for determining the position of honeycomb artifacts provided by the present invention

[0055] Figure 6 ; A strategy diagram for removing glue joint artifacts in the method for improving the image quality of the optical fiber image transmission bundle system provided by the present invention;

[0056] Figure 7 A strategy diagram for removing broken beam artifacts in the method for improving image quality based on the optical fiber image transmission bundle system provided by the present invention. DETAILED DESCRIPTION

[0057] In order to make the objects, advantages and techniques of the present invention clearer and more specific, further detailed description is given below in conjunction with specific implementation modes of the present invention and corresponding drawings.

[0058] Figure 1 This is a system block diagram of a method for improving the image quality of an optical fiber image transmission bundle system according to the present invention. The sensor used is a CCD camera, and other hardware includes an optical fiber bundle and an eyepiece at the front end of the optical fiber bundle. The steps include adding a mask, removing honeycomb artifacts, removing glue artifacts, and removing broken beam artifacts. The specific implementation methods are as follows:

[0059] S1: Adding broken beam and glue mask to the image: Adding broken beam mask to the input image using the filtering method based on binary image, adding glue mask to the input image using the threshold segmentation method. The mask marks the coordinate position of the glue artifact or broken beam artifact in the image, so as to distinguish the honeycomb artifact, glue artifact and broken beam artifact.

[0060] S2: Remove honeycomb artifacts: Use the template to convolve the pixel area outside the glue mask and the broken beam mask in the image and calculate the average value. Set the threshold, determine the location of the honeycomb artifact based on the relationship between the average value and the threshold, and then use the edge arithmetic difference method to remove the honeycomb artifacts to obtain a honeycomb artifact-free image. For the removal of glue and broken beams, the honeycomb artifact-free image is less disturbed by honeycomb artifacts than the original image edge, so it is easier to process;

[0061] S3: Remove the adhesive artifact: Use the bidirectional traversal maximum filtering method to process the adhesive artifact. Calculate the horizontal and vertical coordinates of the two points with the farthest mask distance to determine the traversal method, and then use the bidirectional traversal maximum filtering method to assign the maximum value of the pixels in the neighborhood range to the pixel at the center position of the filter in turn. Calculate the center point of a single row (or column) until the position of the bidirectional traversal reaches the center point, then stop the traversal of this row (or column), and then perform the traversal of the next row (or column).

[0062] S4: Remove the broken beam artifact: Use the bilateral equal difference assignment method to process the broken beam artifact. Take blank areas with a width of four pixels on both the left and right sides of the broken beam mask, and assign values to the broken beam mask area based on the equal difference pixel values shown in the blank areas.

[0063] Figure 2 This is the strategy diagram for adding a mask in a method for improving the image quality of an optical fiber image bundle system according to the present invention, that is, the content of step S1. The specific implementation is as follows:

[0064] S11: Use the optical fiber image bundle system to collect a pure white background image P1;

[0065] S12: Set thresholds V1, V2 (V2 < V1), and C; V1 is used for image binarization, and V2 is used to distinguish the dark adhesive artifacts from the light honeycomb artifacts in the pure white background image P1;

[0066] S13: Binarize the pure white background image P1 based on the threshold V1, and output a binary image P2 with the same size as the original image;

[0067] S14: Calculate and output the number of elements with a value of 1 in the template;

[0068] S15: Use a template similar to the Figure 3 template (the area with elements of 1 has the same pixel width as the core part) to perform convolution on the binary image P2 to obtain the convolution sum;

[0069] S16: Divide the convolution sum obtained in S14 by the number output in S15 to get the average value a;

[0070] S17: Add a broken beam mask to the area where a is less than the threshold C;

[0071] S18: Traverse the original image with the broken beam mask added;

[0072] S19: Add an adhesive mask to the pixel area (outside the broken beam mask area) in the pure white background image P1 that is less than V2;

[0073] Figure 4 This is the strategy diagram for removing honeycomb artifacts in a method for improving the image quality of an optical fiber image bundle system according to the present invention, that is, the content described in S2. The specific implementation steps are as follows:

[0074] S21: Calculate and output the number of elements with a value of 1 in a template whose area of elements with a value of 0 is the same as the pixel width of the core part. Specifically, for example, Figure 5 (the template shown in (1)).

[0075] S22: Traverse the binary image P2 output in S13 using the template and skip the mask area. Specifically, set the coordinates (5, 5) as the center of the template and output the convolution value.

[0076] S23: Divide the convolution sum obtained in S21 by the number output in S22 to get the average value t.

[0077] S24: Set the threshold T. Specifically, T = 20.

[0078] S25: If t < T, then determine that the edge of this area is an artifact, and use the edge arithmetic progression method to remove the artifact at the same position in the original image.

[0079] S26: Calculate and output the number of elements with a value of 1 in a template that is one circle larger than the template in S21. Specifically, for example, Figure 5 (the template shown in (2)).

[0080] S27: Repeat the same method as S22 - S25 to process the image.

[0081] Figure 5 (1) and Figure 5 (2) The positions of the elements with a value of 1 in the templates correspond to the artifact areas in the original image. To make their assignments reasonable and the edges transition naturally and smoothly, the edge arithmetic progression method is used to process the artifact areas. The specific implementation steps are as follows:

[0082] a. Determine the arithmetic progression processing direction. When there are only two elements with a value of 1 in the same row or column (for example, all elements in the template shown in (1) meet this condition), then judge its direction as horizontal or vertical, and the specific orientation is the direction of the 0 element in its adjacent cross area; there is a situation where there are two 0 elements in the cross neighborhood of the element (-3, -4) in the template shown in (1), and uniformly inward, that is, the direction closer to the center of the template is the arithmetic progression processing direction; there is a situation where there are three 0 elements in the neighborhood of the element (-4, -4) in the template shown in (2), and uniformly take the direction closer to the center of the template as the arithmetic progression processing direction; there is a situation where there are no 0 elements in the neighborhood of the element (-5, -5) in the template shown in (2), and uniformly take the direction inward along the line connecting to the center of the template as the arithmetic progression processing direction. Figure 5 (1) All elements in the template shown in (1) meet this condition. Figure 5 (1) In the cross neighborhood of the element (-3, -4) in the template shown in (1), there are two 0 elements. Figure 5 (2) In the neighborhood of the element (-4, -4) in the template shown in (2), there are three 0 elements. Figure 5 (2) In the neighborhood of the element (-5, -5) in the template shown in (2), there are no 0 elements.

[0083] b. Calculate the difference d1 between the pixels in the original image corresponding to the two 0 elements in the template along the processing direction.

[0084] c. Use the pixel with the element 0 in the template next to the point to be processed and add the pixel difference d1 to assign it to the point to be processed, and then complete the processing of one point. For example, the processing method of (-3, -4) in template (2) is:

[0085] d1=P(-3,-3)-P(-3,-2) (1)

[0086] P(-3,-4)=P(-3,-3)+d1 (2)

[0087] Where P(x,y) is the pixel value of the original image corresponding to the coordinate (x,y) in the template, and d1 is the difference between the two pixels.

[0088] For the elements at the corners of template (3), there are no elements around them that are zero. The pixels that meet the processing conditions in the positive direction are processed first, and then these pixels are used to process the pixels at the corners of the corresponding template. For example, for the pixel corresponding to (-5, -5) in template (3), first add the pixel difference d1 to the pixel corresponding to (-3, -3) in the template, and assign it to the pixel corresponding to (-4, -4). Then, add d1 to the pixel corresponding to (-4, -4), and assign it to the pixel corresponding to (-5, -5).

[0089] Figure 6 This is a strategy diagram for removing glue artifacts in step S3 of a method for improving image quality based on an optical fiber image transmission bundle system of the present invention. The specific implementation steps are as follows:

[0090] S31: Calculate the maximum horizontal coordinate difference and the maximum vertical coordinate difference of the glued mask area respectively, that is:

[0091] Δx=x max -x min (3)

[0092] Δy=y max -y min (4)

[0093] Where Δx (or Δy) is the difference between pixels, x max (or y max ) is the maximum value of the horizontal coordinate (or vertical coordinate) in the mask area, x min (or y min ) is the minimum value of the horizontal axis (or vertical axis) in the mask area.

[0094] S32: If Δx ≥ Δy, traverse vertically;

[0095] S33: if the number of mask pixels in a column is an even number, then divide by 2 to obtain the coordinates of the center point; if the number of mask pixels in a column is an odd number, then add 1 and divide by 2 to obtain the coordinates of the center point;

[0096] S34: Use bidirectional traversal, traverse from the leftmost and rightmost sides of this row toward the center point at the same time, stop at the center point on the left side and stop at the point before the center point on the right side to prevent repeated traversal, and proceed to the next row after the traversal is completed.

[0097] S35: If Δx<Δy, then traverse horizontally:

[0098] S36: if the number of mask pixels in a row is an even number, then divide by 2 to obtain the coordinates of the center point; if the number of mask pixels in a row is an odd number, then add 1 and divide by 2 to obtain the coordinates of the center point;

[0099] S37 adopts bidirectional traversal, traversing from the leftmost and rightmost sides of the row toward the center point at the same time. The left side traverses to the center point and stops at the point before the center point, to prevent repeated traversal. After the traversal is completed, proceed to the next row.

[0100] S38: Use maximum value filtering to process the image while traversing: When traversing to (x0, y0), search for the point P(i0, j0) with the largest pixel value in the 3*3 neighborhood and assign its pixel value to the coordinate (x0, y0), as shown in the following formula:

[0101]

[0102] Among them, P′(x0,y0) is the pixel value at (x0,y0).

[0103] Figure 7 This is a strategy diagram of a broken beam removal module based on an image quality improvement method for an optical fiber image transmission bundle system of the present invention, that is, the content described in S4, and the specific implementation steps are as follows:

[0104] S41: taking blank areas with a width of 4 pixels on both sides of the broken beam mask, and the broken beam mask area and the blank area are the same processing area;

[0105] S42: Calculate the difference between the leftmost pixel value and the left and right pixel values ​​of each row in this area;

[0106] S43: Divide the difference by 17 (the number of pixels in this area minus 1) to obtain the tolerance d2;

[0107] S44: Assign values ​​to each pixel in the same row. The assignment method is as follows:

[0108] P(x i ,y j )=[P(x0,y j )+nd2],i=n=1,2,3,…,16 (6)

[0109] In the formula, P(x i ,yj ) is the point to be assigned, P(x0,y j ) is the leftmost point in the row, and operator [x] is the largest integer not exceeding x.

[0110] The above contents are further detailed descriptions of the present invention in combination with specific preferred implementations, and it cannot be determined that the specific implementation of the present invention is limited to these descriptions. For ordinary technicians in the technical field to which the present invention belongs, several simple deductions or substitutions can be made without departing from the concept of the present invention, which should be regarded as falling within the protection scope of the present invention.

Claims

1. A method for improving image quality based on an optical fiber image transmission bundle system, characterized in that: It includes the following steps: S1. Add a broken fiber and glued mask to the image: Process the acquired image and add a mask to the part of the image artifact formed due to broken fibers and gluing; S2. Remove honeycomb artifacts: Determine the position of honeycomb artifacts based on the image with the added mask, and use the edge equal difference method to remove honeycomb artifacts; S3. Remove glued artifacts: Use the maximum filtering method of bidirectional traversal to remove glued artifacts; S4. Remove broken fiber artifacts: Use the bilateral equal difference assignment method to remove broken fiber artifacts.

2. The method for improving image quality based on an optical fiber image transmission bundle system according to claim 1, characterized in that: The step of adding a mask in step S1 includes the following steps: S11: Use the fiber optic image bundle system to acquire a pure white background image P1; S12: Set thresholds V1, V2 (V2 < V1), and C; S13: Binarize the pure white background image P1 based on the threshold V1, and output a binary image P2 with the same size as the original image; S14: Calculate and output the number of elements with a value of 1 in the template; S15: Convolve the binary image P2 with the template to obtain a convolution sum; S16: Divide the convolution sum obtained in S14 by the number output in S15 to get the average value a; S17: Add a broken fiber mask to the area where a is less than the threshold C; S18: Traverse the original image with the added broken fiber mask; S19: Add a glued mask to the pixel area (outside the broken fiber mask area) in the pure white background image P1 that is less than V2.

3. The method for improving image quality based on an optical fiber image transmission bundle system according to claim 1, characterized in that: The step of determining the position of honeycomb artifacts based on the image with the added mask in step S2 includes the following steps: S21: Calculate and output the number of elements with a value of 1 in the template with the same pixel width as the core part; S22: Convolve the binary image P2 output in S13 (skipping the mask area) with the template to obtain a convolution sum; S23: Divide the convolution sum obtained in S22 by the number output in S23 to get the average value t; S24: Set the threshold T; S25: If t < T, then judge that the edge of this area is a honeycomb artifact, and use the edge equal difference method to remove the honeycomb artifact at the same position in the original image; S26: Calculate and output the number of elements with a value of 1 in the template that is one circle larger than that in S21; S27: Repeat the same method as S22 - S25 to process the image.

4. The method for improving image quality based on an optical fiber image transmission bundle system according to claim 1, characterized in that: The step of removing honeycomb artifacts by the edge equal difference method in step S2 includes the following steps: a. Determine the equal difference processing direction: When there are only two elements with a value of 1 in the same row or column, then judge its direction as horizontal or vertical, and the specific orientation is the direction of 0 in its adjacent cross area; when there are two 0 elements in the cross neighborhood, uniformly inward; when there are three 0 elements in the neighborhood, uniformly take the direction closer to the center of the template as the equal difference processing direction; when there are no 0 elements in the neighborhood, uniformly take the direction inward along the line connecting to the center of the template as the equal difference processing direction; b. Calculate the difference d1 between the pixels in the original image corresponding to the two elements with a value of 0 in the template along the processing direction; c. Use the pixel adjacent to the point to be processed corresponding to the element with a value of 0 in the template plus the pixel difference d1, and assign it to the point to be processed.

5. The method for improving image quality based on an optical fiber image transmission bundle system according to claim 1, characterized in that: The step of removing glued artifacts in step S3 includes the following steps: S31: Calculate the difference between the maximum abscissa and the maximum ordinate Δx and Δy of the mask area respectively; S32: If Δx ≥ Δy, then traverse vertically; S33: Set the center point of each column; S34: traverse from the leftmost and rightmost sides of this column towards the center point at the same time, traverse in both directions until the center point and stop traversing, and then proceed to the next column after completion; S35: if Δx<Δy, traverse horizontally; S36: Set the center point of each row; S37: traverse from the leftmost and rightmost sides of this row towards the center point at the same time, traverse in both directions until the center point and stop traversing, and then proceed to the next row after completion; S38: Search for the point with the largest pixel in the 3*3 neighborhood and assign its value to the point to be processed.

6. The method for improving image quality based on an optical fiber image transmission bundle system according to claim 1, characterized in that: The removal of beam break artifacts in step S4 comprises the following steps: S41: taking a blank area with a certain pixel width on both sides of the broken beam mask, and making the blank area and the broken beam mask area into a processing area; S42: Calculate the difference between the leftmost pixel value and the left and right pixel values ​​of each row in this area; S43: Divide the difference by the distance between the two farthest pixels to obtain a tolerance d2; S44: Perform arithmetical assignment of values ​​to pixels in the same row.

Images