Three-dimensional recursive filtering method and device for image time-space domain combined noise reduction

Abstract

The invention provides a three-dimensional recursive filtering method and device for image time-space domain combined noise reduction. The image time-space domain combined noise reduction method is different from the thought that traditional three-dimensional recursive filtering is divided into three one-dimensional filtering bodies which are directly connected in series, the time domain recursive information is fused in the point-by-point space domain recursive process, the space domain filtering effect is improved, then the space domain recursive information is fused in the point-by-point time domain filtering process, and the noise reduction effect is improved. Compared with a traditional time-space domain series connection method and a time-space domain parallel connection method, the method has the advantages that the image definition and the signal-to-noise ratio can be better improved, and the phenomena of lack of motion smear and raindrop noise are reduced at the same time under the operation logic and storage overhead of the same level.

Description

Technical Field

[0001] The present invention relates to the field of image processing technology, and in particular to a three-dimensional recursive filtering method and device for joint denoising of images in time and space domains. Background Art

[0002] One of the important links in image processing technology is to reduce the noise of images to obtain cleaner and clearer images. It is usually assumed that additive random noise will be generated when the image is collected by the sensor, and the noise mean is 0 to perform weighted removal. For video noise reduction, there are two main methods, namely time domain noise reduction and spatial domain noise reduction. Time domain noise reduction uses the correlation of the time domain and historical information, such as the image content of the previous frame, to reduce the noise of the static area of ​​the image of the current frame. Spatial domain noise reduction uses the local spatial similar image content of the current image to reduce the noise. The other is to use domain transformations such as discrete cosine transform and discrete wavelet transform to filter different frequency domains and then convert them back to reduce the noise of the image.

[0003] Temporal noise reduction can denoise static areas while improving their clarity, but it cannot denoise moving areas. In addition, the trade-off between motion smear loss and rain noise is difficult to handle. Either motion smear loss is caused in order to suppress rain noise, or rain noise is caused in order to avoid motion smear and loss.

[0004] Although the denoising strength of the spatial domain denoising method will become stronger as the receptive field increases, there is always a trade-off between denoising strength and clarity. Either the spatial domain denoising strength is weakened to maintain clarity, or some clarity is sacrificed to improve the image signal-to-noise ratio.

[0005] Spatiotemporal denoising usually adopts a combination of methods. For example, temporal denoising is performed first, and then spatial denoising of different intensities is applied to the moving area and static area of ​​the temporal denoising result in series. Or spatial denoising is performed first, and then temporal denoising is applied to the result of spatial denoising in series. Or temporal denoising is performed on the original image at the same time as spatial denoising, and the results are weighted in parallel.

[0006] Usually the effect of series connection is better than that of parallel connection. This is because the previous level of denoising in series will improve the signal-to-noise ratio of the image, thereby reducing the difficulty of the next level of denoising. For example, after the spatial domain denoising is pre-processed with time domain denoising, the signal-to-noise ratio of the static area is improved, and the clarity is improved while the noise is reduced. At this time, spatial domain denoising can better distinguish between texture and noise, thereby solving the problem of the trade-off between clarity and noise reduction. Similarly, after the spatial domain denoising is pre-processed with time domain denoising, the signal-to-noise ratio of the current image is improved. Since raindrop noise is caused by noise being judged as motion, the noise is reduced, thereby reducing the raindrop noise phenomenon, and the motion is more obvious, which can better solve the problem of the trade-off between motion smear and raindrop noise. However, in the series structure, one of the two is destined to be placed in front, that is, the problem of the filter itself placed in front can never be solved.

[0007] A recursive filter is an IIR filter, also known as a feedback filter. Similar to traditional filtering methods, the output of the filter depends on the input and filter response function like traditional filtering, but the difference is that recursive filtering depends on past outputs. Based on the recursive structure, complex recursive filters such as Kalman filter, particle filter, Gaussian recursive filter, etc. are designed. One-dimensional recursive filtering is used for time domain noise reduction, two-dimensional recursive filtering is used for spatial domain noise reduction, and three-dimensional recursive filtering is used for time and space domain noise reduction. For high-dimensional recursive filtering, it is usually disassembled into a series structure of multiple one-dimensional filters. For example, for two-dimensional recursive filtering, a one-dimensional recursive filter in the horizontal direction is performed first, and then a one-dimensional recursive filter in the vertical direction. For three-dimensional recursive filtering, a one-dimensional recursive filter in the time domain is added in series on the basis of the two-dimensional recursive filter. However, this disassembly method still has the problem that the problem of the previous stage of the time and space domain series cannot be alleviated. Summary of the invention

[0008] The purpose of the present invention is to provide a three-dimensional recursive filtering method and device for joint denoising of images in time and space domains, so as to solve the problem that the filtering problem of the former filtering in the traditional series structure of time and space domain filtering is not improved by the series connection.

[0009] To solve the above technical problems, based on one aspect of the present invention, the present invention provides a three-dimensional recursive filtering method for joint denoising of images in spatiotemporal and temporal domains, which comprises:

[0010] Step 1: Determine the point to be processed of the current image and the reference point of the reference image at the same position according to the recursive filtering direction;

[0011] Step 2: performing spatial domain recursive filtering on the difference between the to-be-processed point of the current image and the reference point of the reference image to obtain a difference filtering result and a spatial domain filtering coefficient;

[0012] Step 3: performing time-domain recursive filtering on the points to be processed of the current image according to the difference filtering result to obtain updated points to be processed of the current image;

[0013] Step 4: Calculate the directional gradient of the to-be-processed point of the current image and the directional gradient of the reference point of the reference image;

[0014] Step 5: Based on the spatial domain filtering coefficient, the directional gradient of the to-be-processed point of the current image is subjected to time domain recursive filtering, thereby obtaining a time domain filtering result;

[0015] Step 6: Based on the time domain filtering result, perform spatial domain recursive filtering on the updated current image to-be-processed points to obtain the current spatial domain filtering result;

[0016] Step 7: performing spatial domain recursive filtering on the reference point of the reference image according to the directional gradient of the reference point of the reference image to obtain a reference spatial domain filtering result;

[0017] Step eight: Update the difference filtering result according to the current spatial domain filtering result and the reference spatial domain filtering result.

[0018] Optionally, the method further includes:

[0019] Step nine: determine whether all the points to be processed in the current image in the recursive filtering direction have completed steps one to eight; if so, execute step ten; if not, find the next point to be processed in the current image and execute steps one to eight;

[0020] Step 10: According to the updated difference filtering results of all the to-be-processed points of the current image, the current image is subjected to time-domain recursive filtering processing, and a final filtering result is output.

[0021] Optionally, the reference image is an image of a previous frame of the current image that has output the final filtering result.

[0022] Based on another aspect of the present invention, the present invention also provides a three-dimensional recursive filtering device for joint denoising of images in time and space domains, which comprises:

[0023] A data input unit, used to obtain a current image and a reference image, and determine a to-be-processed point of the current image and a reference point of the reference image at the same position according to a recursive filtering direction;

[0024] An image difference image gradient calculation unit, used to calculate the difference between the to-be-processed point of the current image and the reference point of the reference image, and to calculate the directional gradient of the to-be-processed point of the current image and the directional gradient of the reference point of the reference image;

[0025] An image difference spatial domain recursive filtering unit, used for performing spatial domain recursive filtering on the difference between the to-be-processed point of the current image and the reference point of the reference image to obtain a difference filtering result and a spatial domain filtering coefficient;

[0026] An image directional gradient time-domain recursive filtering unit, used to perform time-domain recursive filtering on the directional gradient of the to-be-processed point of the current image based on the spatial domain filtering coefficient, thereby obtaining a time-domain filtering result;

[0027] An image time-domain recursive unit, configured to perform time-domain recursive filtering on the to-be-processed points of the current image according to the difference filtering result to obtain updated to-be-processed points of the current image;

[0028] An image spatial domain recursive filtering unit, configured to perform spatial domain recursive filtering on the updated current image to-be-processed point based on the temporal domain filtering result to obtain a current spatial domain filtering result, and perform spatial domain recursive filtering on the reference point of the reference image according to the directional gradient of the reference point of the reference image to obtain a reference spatial domain filtering result;

[0029] An image difference updating unit is used to update the difference filtering result according to the current spatial domain filtering result and the reference spatial domain filtering result.

[0030] Optionally, the device also includes a judgment unit and a time domain filtering post-processing unit, the judgment unit being used to judge whether all the points to be processed in the current image in the recursive filtering direction have updated the difference filtering results. If so, the time domain filtering post-processing unit performs time domain recursive filtering on the current image based on the updated difference filtering results of all the points to be processed in the current image, and outputs the final filtering result. If not, the judgment unit searches for the next point to be processed in the current image and inputs it to the data input unit.

[0031] As mentioned above, the three-dimensional recursive filtering method for joint spatiotemporal domain denoising of images of the present invention is different from the idea of ​​splitting the traditional three-dimensional recursive filtering into three one-dimensional filters that are directly connected in series. Instead, it improves the effect of spatial filtering by integrating the information of time domain recursion into the process of point-by-point spatial recursion, and then improves the effect of time domain filtering by integrating the information of spatial domain recursion into the process of point-by-point time domain filtering. The entire image is repeatedly recursively traversed in this way. Under the same level of computing logic and storage overhead, the present invention can better improve image clarity and signal-to-noise ratio than traditional spatiotemporal domain series and spatiotemporal domain parallel methods, while reducing the lack of motion ghosting and raindrop noise.

[0032] It should be noted that the three-dimensional recursive filtering device for joint denoising of images in spatiotemporal domains and the three-dimensional recursive filtering method for joint denoising of images in spatiotemporal domains are based on the same inventive concept and have the same or corresponding specific technical features. Therefore, the three-dimensional recursive filtering device for joint denoising of images in spatiotemporal domains also has the technical effects of the three-dimensional recursive filtering method for joint denoising of images in spatiotemporal domains, which will not be repeated here. BRIEF DESCRIPTION OF THE DRAWINGS

[0033] Those skilled in the art should understand that the drawings are provided for a better understanding of the present invention and do not constitute any limitation on the scope of the present invention.

[0034] Figure 1 is a schematic diagram of a three-dimensional recursive filtering method for joint denoising of images in the spatiotemporal domain according to an embodiment of the present invention;

[0035] Figure 2 It is a schematic diagram of a three-dimensional recursive filtering device for joint denoising of images in the spatiotemporal domain according to an embodiment of the present invention. DETAILED DESCRIPTION

[0036] In order to make the purpose, advantages and features of the present invention clearer, the present invention is further described in detail below in conjunction with the accompanying drawings and specific embodiments. It should be noted that the drawings are all in a very simplified form and are not drawn to scale, and are only used to conveniently and clearly assist in explaining the purpose of the embodiments of the present invention. In addition, the structure shown in the drawings is often a part of the actual structure. In particular, the emphasis of each drawing is different, and sometimes different scales are used.

[0037] As used in the present invention, the singular forms "one", "an" and "the" include plural objects, the term "or" is generally used to include the meaning of "and / or", the term "several" is generally used to include the meaning of "at least one", and the term "at least two" is generally used to include the meaning of "two or more". In addition, the terms "first", "second" and "third" are only used for descriptive purposes and cannot be understood as indicating or implying relative importance or implicitly indicating the number of technical features indicated. Therefore, the features defined as "first", "second" and "third" may explicitly or implicitly include one or at least two of the features, "one end" and "the other end" and "the proximal end" and "the distal end" generally refer to two corresponding parts, which include not only the endpoints, and the terms "installed", "connected" and "connected" should be understood in a broad sense, for example, it can be a fixed connection, a detachable connection, or an integral body; it can be a mechanical connection or an electrical connection; it can be directly connected or indirectly connected through an intermediate medium, it can be the internal connection of two elements or the interaction relationship between two elements. In addition, as used in the present invention, an element disposed on another element generally only indicates that there is a connection, coupling, cooperation or transmission relationship between the two elements, and the connection, coupling, cooperation or transmission between the two elements may be direct or indirect through an intermediate element, and cannot be understood as indicating or implying the spatial position relationship between the two elements, that is, one element may be in any orientation such as inside, outside, above, below or on one side of another element, unless otherwise clearly indicated in the content. For ordinary technicians in this field, the specific meanings of the above terms in the present invention can be understood according to specific circumstances.

[0038] Figure 1 FIG. 1 is a schematic diagram of a three-dimensional recursive filtering method for joint denoising of images in the temporal and spatial domains according to an embodiment of the present invention. Figure 1 An embodiment of the present invention provides a three-dimensional recursive filtering method for joint denoising of images in time and space domains. The method is used to perform three-dimensional recursive filtering on an image. Specifically, a current image and a reference image are first obtained, the current image is traversed, and the image is subjected to three-dimensional recursive filtering through the following steps 1 to 10.

[0039] Step 1: Determine the processing point C (h1, w1, t1) of the current image and the reference point P (h1, w1, t0) of the reference image at the same position according to the recursive filtering direction. It can be seen that the same position here means that the positions of the processing points of the current image and the reference image are the same. It can be understood that the processing points of the image refer to the pixel points that need special attention or processing during the image processing process. These points may be noise points, edge points, feature points, etc. in the image, and specific processing is required to improve the quality of the image or extract useful information.

[0040] Step 2: Calculate the difference between the processing point C(h1, w1, t1) of the current image and the reference point P(h1, w1, t0) of the reference image to obtain the difference Dif(h1, w1, t1), and perform spatial recursive filtering on the difference Dif(h1, w1, t1) to obtain the difference filtering result Dif_r(h1, w1, t1) and the spatial filtering coefficient.

[0041] Step three: Perform time-domain recursive filtering on the pending point C(h1, w1, t1) of the current image according to the difference filtering result Dif_r(h1, w1, t1) to obtain the updated pending point C_t(h1, w1, t1) of the current image.

[0042] Step 4: Calculate the directional gradient Gc(h1, w1, t1) of the point to be processed in the current image and the directional gradient Gp(h1, w1, t0) of the reference point in the reference image.

[0043] Step 5: Based on the spatial domain filtering coefficient, the directional gradient Gc(h1, w1, t1) of the point to be processed of the current image is subjected to time domain recursive filtering, thereby obtaining a time domain filtering result Gc_t(h1, w1, t1).

[0044] Step six: Based on the time domain filtering result Gc_t(h1, w1, t1), the updated current image processing point C_t(h1, w1, t1) is subjected to spatial domain recursive filtering to obtain the current spatial domain filtering result C_tr(h1, w1, t1).

[0045] Step seven: Perform spatial domain recursive filtering on the reference point P (h1, w1, t0) of the reference image according to the directional gradient Gp (h1, w1, t0) of the reference point of the reference image to obtain a reference spatial domain filtering result P_r (h1, w1, t0).

[0046] Step eight: Update the difference filtering result according to the current spatial filtering result C_tr(h1, w1, t1) and the reference spatial filtering result P_r(h1, w1, t0), so as to obtain an updated difference filtering result Dif_rt(h1, w1, t1).

[0047] Step nine: Determine whether all the points to be processed in the recursive filtering direction of the current image have completed steps one to eight, that is, determine whether all the points to be processed in the recursive filtering direction of the current image have all been iterated three-dimensionally recursively filtered and updated difference filtering results have been obtained; if so, execute step ten; if not, find the next point to be processed in the recursive filtering direction of the current image, and execute steps one to eight, so as to perform three-dimensional recursive filtering on the next point to be processed.

[0048] Step ten: According to the updated difference filtering results of all the to-be-processed points of the current image, the current image is subjected to time-domain recursive filtering processing, and the final filtering result is output, thereby realizing three-dimensional recursive filtering processing of the current image.

[0049] It should be noted that the reference image is generally selected as the image of the previous frame of the current image that has output the final filtering result, that is, the reference image is selected as the image of the previous frame that has been completely iterated to complete the three-dimensional recursive filtering process of the present invention. The reference image is selected in this way, and the latest updated image can be used as the reference image of the current image, reducing the difference between the current image and the reference image, which is conducive to improving the filtering processing effect and improving the quality of image processing.

[0050] In this way, when the current image, the reference image, and the points to be processed of the image difference are updated, the image gradient is filtered in the time domain using the spatial filtering information of the image difference, and then the image gradient after the time domain filtering is used to filter the current image in the space domain, and finally the frame difference is filtered and updated according to the filtered current image and the reference image. The time domain and space domain filtering are repeatedly and alternately executed to continuously improve the signal-to-noise ratio of the processed information. On this basis, the recursive filtering itself will recurse the above steps multiple times in different directions for each processing point. Finally, an output result with a very high signal-to-noise ratio can be output, which can effectively handle scenes with a very poor signal-to-noise ratio. The three-dimensional recursive filtering method for joint spatiotemporal domain denoising of images of the present invention is different from the idea of ​​splitting the traditional three-dimensional recursive filtering into three one-dimensional filters that are directly connected in series. Instead, it improves the effect of spatial filtering by integrating the information of time domain recursion in the process of point-by-point spatial recursion, and then improves the effect of time domain filtering by integrating the information of spatial domain recursion in the process of point-by-point time domain filtering. The entire image is repeatedly recursively traversed in this way. Under the same level of computing logic and storage overhead, the present invention can better improve image clarity and signal-to-noise ratio than traditional spatiotemporal domain series and spatiotemporal domain parallel methods, while reducing the lack of motion ghosting and raindrop noise.

[0051] It can be understood that the three-dimensional recursive filtering process as described above is performed on the current image, including the processing process in the horizontal direction and the vertical direction, that is, the recursive filtering direction selects the horizontal direction or the vertical direction. The logical steps of different recursive filtering directions are the same, except that the reference points of the selected reference images are different. The following is an example of forward recursive filtering in the horizontal direction. Assume that the current image to be filtered is C0, the reference image to be filtered is P0, and the image difference to be filtered is Dif0. C1, P1, and Dif1 are the results after filtering C0, P0, and Dif0, respectively.

[0052] First, after determining the processing point C0(i, j) of the current image and the reference point P0(i, j) of the reference image, the calculation formula of the difference Dif0 between the processing point of the current image and the reference point of the reference image is as follows:

[0053] Dif0(i,j)=C0(i,j)-P0(i,j)

[0054] To perform spatial filtering, it is necessary to determine the reference point of the filter. Since it is a horizontal forward recursive filter, for the position (i, j) of the current point to be filtered, the value at the position (i, j-1) is the output result of the horizontal forward recursive filter. Therefore, the reference points are C1 (i, j-1), P1 (i, j-1) and Dif1 (i, j-1).

[0055] The spatial domain filter coefficient Wd of the difference between the current image and the reference image is calculated by referring to the following formula:

[0056]

[0057] where sigma d is the filtering strength. Then the spatial filtering result Dif0 of the difference between the current image and the reference image at the current position to be processed R for:

[0058] Dif0 R =Dif0(i,j)*(1-Wd)+Dif1(i,j-1)*Wd

[0059] Calculate the gradients of the current image and the reference image in four directions. The gradient calculation uses the convolution kernel, and the convolution kernel uses the Sobel operator, as follows:

[0060] Wg0=[-1 1 0]

[0061] Wg1=[0 1-1]

[0062]

[0063] This example is a horizontal forward filter, and Wg0 is selected to calculate the gradient, that is, to calculate the left gradient. The other three cases use the last three convolution kernels to calculate the gradient. At this time, the calculation formulas for the left gradients Gc0 and Gp0 of the current image and the reference image are as follows:

[0064] Gc0=C0(i,j)-C1(i,j-1)

[0065] Gp0=P0(i,j)-P1(i,j-1)

[0066] It is not difficult to see the relationship between the left gradient and image difference between the current image and the reference image through the following recursive formula:

[0067] Dif0(i,j)-Dif1(i,j-1)

[0068] =(C0(i,j)-P0(i,j))-(C1(i,j-1)-P1(i,j-1))

[0069] =(C0(i,j)-C1(i,j-1))-(P0(i,j)-P1(i,j-1))

[0070] =Gc0-Gp0

[0071] Therefore, the temporal filtering coefficient Wg of the temporal filtering of the image gradient is equal to the spatial filtering coefficient Wd of the difference between the current image and the reference image. Then the temporal filtering result Gc0 of the left direction gradient of the current image at the current position to be processed is T for:

[0072] Gc0 T =Gc0(i,j)*(1-Wg)+Gc1(i,j-1)*Wg

[0073] Among them, Wg=Wd.

[0074] According to the output Dif0 R , calculate the time domain filtering result C0 of the current image at the current position to be processed T , the formula is as follows:

[0075]

[0076] C0 T =C0(i,j)*(1-Wct)+P0(i,j)*Wct

[0077] where sigma ct is the filter strength.

[0078] Then according to the output Gc0 T , calculate the spatial filtering result C1(i,j) of the current image at the current position to be processed, the formula is as follows:

[0079]

[0080] C1(i,j)=C0T*(1-Wcr)+C1(i,j-1)*Wcr

[0081] where sigma cr is the filter strength.

[0082] According to Gp0, calculate the spatial filtering result P1(i,j) of the reference image at the current position to be processed, the formula is as follows:

[0083]

[0084] P1(i,j)=P0(i,j)*(1-Wpr)+P1(i,j-1)*Wpr

[0085] Among them, sigma Pr is the filter strength.

[0086] According to C1(i, j) and P1(i, j), the difference Dif1(i, j) between the filtered current image and the reference image at this position is calculated. The formula is as follows:

[0087] Dif1(i,j)=C1(i,j)-P1(i,j)

[0088] Let j=j+1 and repeat the above steps and logic.

[0089] The process operation logic of other recursive directions is the same as the above example, except that the recursive direction and input and output data are different. It is not difficult to see that the final output Dif_final and C_final are the results of recursive generation of relevant information that continuously refers to the time domain and spatial domain updates, and P_final is obtained by recursive filtering of the reference image P0 in pure spatial domain. To ensure the final clarity, the spatial domain recursive filtering strength of the reference image should be weaker than the spatial domain filtering strength of the current image. Once the time domain filtering makes a lack of motion, the subsequent recursion cannot make up for the lack of motion, and the same point will be recursively filtered multiple times in different directions. Therefore, the intensity of the time domain filtering in step three of the recursive filtering of C_final should be as weak as possible. Considering that the time domain filtering strength in different direction recursions is weak, and each time domain filtering uses the image difference of the non-final filtering result to calculate the filter coefficient, it is necessary to use the final filtering result of the image difference to calculate the time domain filter coefficient that is cleaner, has more complete motion and has a larger filtering strength as the end, so as to perform the final time domain recursive filtering on the current image, thereby outputting the final filtering result. The following is the time domain filter coefficient W final The calculation formula is:

[0090]

[0091] where sigma final is the final time domain filtering strength.

[0092] The final filtering result output calculation formula after joint filtering in the time and space domains is as follows:

[0093] output=C_final*(1-W final )+P_final*W final

[0094] Example explanation, here final is 3, according to the above steps logic, first perform horizontal forward filtering on C0, P0 and Dif0 to obtain C1, P1 and Dif1, then perform horizontal reverse filtering on C1, P1 and Dif1 to obtain C2, P2 and Dif2, and then perform vertical forward filtering on C2, P2 and Dif2 to obtain C3, P3 and Dif3.

[0095] Figure 2 Schematic diagram of a three-dimensional recursive filtering device for image spatiotemporal domain joint denoising according to an embodiment of the present invention. Based on the same inventive concept as the above-mentioned three-dimensional recursive filtering method for image spatiotemporal domain joint denoising, the present invention also provides a three-dimensional recursive filtering device for image spatiotemporal domain joint denoising, see Figure 2 , the device comprises:

[0096] A data input unit, used to obtain a current image and a reference image, and determine a to-be-processed point of the current image and a reference point of the reference image at the same position according to a recursive filtering direction;

[0097] An image difference image gradient calculation unit, used to calculate the difference between the to-be-processed point of the current image and the reference point of the reference image, and to calculate the directional gradient of the to-be-processed point of the current image and the directional gradient of the reference point of the reference image;

[0098] An image difference spatial domain recursive filtering unit, used for performing spatial domain recursive filtering on the difference between the to-be-processed point of the current image and the reference point of the reference image to obtain a difference filtering result and a spatial domain filtering coefficient;

[0099] An image directional gradient time-domain recursive filtering unit, used to perform time-domain recursive filtering on the directional gradient of the to-be-processed point of the current image based on the spatial domain filtering coefficient, thereby obtaining a time-domain filtering result;

[0100] An image time-domain recursive unit, configured to perform time-domain recursive filtering on the to-be-processed points of the current image according to the difference filtering result to obtain updated to-be-processed points of the current image;

[0101] An image spatial domain recursive filtering unit, configured to perform spatial domain recursive filtering on the updated current image to-be-processed point based on the temporal domain filtering result to obtain a current spatial domain filtering result, and perform spatial domain recursive filtering on the reference point of the reference image according to the directional gradient of the reference point of the reference image to obtain a reference spatial domain filtering result;

[0102] An image difference updating unit is used to update the difference filtering result according to the current spatial domain filtering result and the reference spatial domain filtering result.

[0103] Furthermore, the device also includes a judgment unit and a time domain filtering post-processing unit, the judgment unit is used to judge whether all the points to be processed in the recursive filtering direction of the current image have updated the difference filtering results. If so, the time domain filtering post-processing unit performs time domain recursive filtering on the current image according to the updated difference filtering results of all the points to be processed in the current image, and outputs the final filtering result. If not, the judgment unit searches for the next point to be processed of the current image and inputs it to the data input unit.

[0104] It should be noted that, those skilled in the art can understand the three-dimensional recursive filtering device for joint denoising in the spatiotemporal domains of images based on the previous description of the three-dimensional recursive filtering method for joint denoising in the spatiotemporal domains of images, as well as the coordination relationship and logical execution relationship between the various units in the device, which will not be elaborated here.

[0105] Although the present invention is disclosed as above with preferred embodiments, the above embodiments are not intended to limit the present invention. For any technician familiar with the art, without departing from the scope of the technical solution of the present invention, the technical content disclosed above can be used to make many possible changes and modifications to the technical solution of the present invention, or modified into equivalent embodiments of equivalent changes. Therefore, any simple modification, equivalent change and modification made to the above embodiments according to the technical essence of the present invention without departing from the content of the technical solution of the present invention still belongs to the scope of protection of the technical solution of the present invention.

Claims

1. A three-dimensional recursive filtering method for joint denoising of images in temporal and spatial domains, characterized in that: include: Step 1: Determine the point to be processed of the current image and the reference point of the reference image at the same position according to the recursive filtering direction; Step 2: performing spatial domain recursive filtering on the difference between the to-be-processed point of the current image and the reference point of the reference image to obtain a difference filtering result and a spatial domain filtering coefficient; Step 3: performing time-domain recursive filtering on the points to be processed of the current image according to the difference filtering result to obtain updated points to be processed of the current image; Step 4: Calculate the directional gradient of the to-be-processed point of the current image and the directional gradient of the reference point of the reference image; Step 5: Based on the spatial domain filtering coefficient, the directional gradient of the to-be-processed point of the current image is subjected to time domain recursive filtering, thereby obtaining a time domain filtering result; Step 6: Based on the time domain filtering result, perform spatial domain recursive filtering on the updated current image to-be-processed points to obtain the current spatial domain filtering result; Step 7: performing spatial domain recursive filtering on the reference point of the reference image according to the directional gradient of the reference point of the reference image to obtain a reference spatial domain filtering result; Step eight: Update the difference filtering result according to the current spatial domain filtering result and the reference spatial domain filtering result.

2. The three-dimensional recursive filtering method for joint denoising in the spatiotemporal domain of images according to claim 1, characterized in that: The method further comprises: Step nine: determine whether all the points to be processed in the current image in the recursive filtering direction have completed steps one to eight; if so, execute step ten; if not, find the next point to be processed in the current image and execute steps one to eight; Step 10: According to the updated difference filtering results of all the to-be-processed points of the current image, the current image is subjected to time-domain recursive filtering processing, and a final filtering result is output.

3. The three-dimensional recursive filtering method for joint denoising in the spatiotemporal domain of images according to claim 2, characterized in that: The reference image is an image of a previous frame of the current image that has output the final filtering result.

4. A three-dimensional recursive filtering device for joint denoising of images in temporal and spatial domains, characterized in that: include: A data input unit, used to obtain a current image and a reference image, and determine a to-be-processed point of the current image and a reference point of the reference image at the same position according to a recursive filtering direction; An image difference image gradient calculation unit, used to calculate the difference between the to-be-processed point of the current image and the reference point of the reference image, and to calculate the directional gradient of the to-be-processed point of the current image and the directional gradient of the reference point of the reference image; An image difference spatial domain recursive filtering unit, used for performing spatial domain recursive filtering on the difference between the to-be-processed point of the current image and the reference point of the reference image to obtain a difference filtering result and a spatial domain filtering coefficient; An image directional gradient time-domain recursive filtering unit, used to perform time-domain recursive filtering on the directional gradient of the to-be-processed point of the current image based on the spatial domain filtering coefficient, thereby obtaining a time-domain filtering result; An image time-domain recursive unit, configured to perform time-domain recursive filtering on the to-be-processed points of the current image according to the difference filtering result to obtain updated to-be-processed points of the current image; An image spatial domain recursive filtering unit, configured to perform spatial domain recursive filtering on the updated current image to-be-processed point based on the temporal domain filtering result to obtain a current spatial domain filtering result, and perform spatial domain recursive filtering on the reference point of the reference image according to the directional gradient of the reference point of the reference image to obtain a reference spatial domain filtering result; An image difference updating unit is used to update the difference filtering result according to the current spatial domain filtering result and the reference spatial domain filtering result.

5. The three-dimensional recursive filtering device for joint denoising in the spatiotemporal and temporal domains of an image according to claim 4, characterized in that: The device also includes a judgment unit and a time domain filtering post-processing unit. The judgment unit is used to judge whether all the points to be processed in the current image in the recursive filtering direction have updated the difference filtering results. If so, the time domain filtering post-processing unit performs time domain recursive filtering on the current image according to the updated difference filtering results of all the points to be processed in the current image, and outputs the final filtering result. If not, the judgment unit searches for the next point to be processed in the current image and inputs it to the data input unit.

Images