WO2022133664A1

WO2022133664A1 - Pixel interpolation method and apparatus, electronic device and computer-readable medium

Info

Publication number: WO2022133664A1
Application number: PCT/CN2020/138064
Authority: WO
Inventors: 唐小东; 王亚强
Original assignee: 深圳市大疆创新科技有限公司
Priority date: 2020-12-21
Filing date: 2020-12-21
Publication date: 2022-06-30

Abstract

The embodiments of the present application disclose a pixel interpolation method and apparatus, an electronic device, and a computer-readable medium. Said method comprises: selecting a target interpolation position of a target image block; querying, from a preset mapping table, a target pixel matrix offset corresponding to the target interpolation position and a target filtering parameter corresponding to the target interpolation position; and performing interpolation on the target interpolation position on the basis of the target pixel matrix offset and the target filtering parameter. The present implementation method improves the pixel interpolation efficiency.

Description

Pixel interpolation method, apparatus, electronic device and computer readable medium

technical field

The embodiments of the present application relate to the field of computer technologies, and in particular, to a pixel interpolation method, an apparatus, an electronic device, and a computer-readable medium.

Background technique

In video, image and other processing scenarios, pixel interpolation is usually required. For example, in the process of video coding, motion estimation needs to be performed in the inter-frame prediction stage, so as to find the image block most similar to the current coding block in the reference frame. Motion estimation includes Integer Motion Estimation (IME) and Fraction Motion Estimation (FME). Integer-pixel motion estimation is to perform a matching search with integer-pixel precision within a certain range of the reference frame to search for the most matching image block with integer-pixel precision. Pixel-by-pixel motion estimation is based on whole-pixel motion estimation to perform matching search with pixel-by-pixel precision to improve the accuracy of motion estimation. Since fractional pixel values do not exist in digital video, the pixel values at these positions need to be calculated by interpolation.

In the prior art, in the process of pixel interpolation, it is usually necessary to determine the interpolation parameters in real time according to the current interpolation position, such as the offset of the pixel matrix required by the interpolation position compared to the input pixel matrix, filtering parameters, etc., so as to determine the interpolation parameters based on the interpolation parameters. Perform sub-pixel interpolation.

The disadvantage of the prior art is that the interpolation parameters are calculated in real time during the interpolation process, resulting in low interpolation efficiency.

SUMMARY OF THE INVENTION

The embodiments of the present application propose a pixel interpolation method, an apparatus, an electronic device, and a computer-readable medium, so as to solve the technical problem of low pixel interpolation efficiency in the prior art.

In a first aspect, an embodiment of the present application provides a pixel interpolation method, including: selecting a target interpolation position of a target image block; querying a target pixel matrix offset corresponding to the target interpolation position and The target filtering parameter corresponding to the target interpolation position; based on the target pixel matrix offset and the target filtering parameter, the target interpolation position is interpolated.

In a second aspect, an embodiment of the present application provides a pixel interpolation device, including: a selection unit, configured to select a target interpolation position of a target image block; a query unit, configured to query the preset mapping table for the the target pixel matrix offset corresponding to the target interpolation position and the target filtering parameter corresponding to the target interpolation position; the interpolation unit is configured to, based on the target pixel matrix offset and the target filtering parameter, perform interpolation on the target position to interpolate.

In a third aspect, embodiments of the present application provide an electronic device, including: a processor and a memory; a memory for storing program instructions; a processor for executing program instructions stored in the memory, and when the program instructions are executed, processing The device is used to perform the following steps: select the target interpolation position of the target image block; from the preset mapping table, query the target pixel matrix offset corresponding to the target interpolation position and the target filtering parameter corresponding to the target interpolation position; The target pixel matrix offset and the target filter parameter are used to interpolate the target interpolation position.

In a fourth aspect, an embodiment of the present application provides a computer-readable medium on which a computer program is stored, and when the program is executed by a processor, implements the above pixel interpolation method.

Embodiments of the present application provide a pixel interpolation method, an apparatus, an electronic device, and a computer-readable medium, which improve pixel interpolation efficiency.

Description of drawings

Other features, objects and advantages of the present application will become more apparent by reading the detailed description of non-limiting embodiments made with reference to the following drawings:

1A is a schematic diagram of a video encoding process;

1B is a schematic diagram of sub-pixel positions;

FIG. 2 is a flowchart of an embodiment of a pixel interpolation method according to the present application;

3 is a schematic diagram of the distribution of interpolation positions according to the pixel interpolation method of the present application;

4 is a schematic diagram of first mapping information according to the pixel interpolation method of the present application;

5 is a schematic diagram of second mapping information according to the pixel interpolation method of the present application;

6 is a schematic diagram of a hardware framework for calculating an interpolation result according to the pixel interpolation method of the present application;

7 is a schematic structural diagram of an embodiment of a pixel interpolation device according to the present application;

FIG. 8 is a schematic structural diagram of an embodiment of an electronic device according to the present application.

specific embodiment

The present application will be further described in detail below with reference to the accompanying drawings and embodiments. It should be understood that the specific embodiments described herein are only used to explain the related invention, but not to limit the invention. In addition, it should be noted that, for the convenience of description, only the parts related to the related invention are shown in the drawings.

It should be noted that the embodiments in the present application and the features of the embodiments may be combined with each other in the case of no conflict. The present application will be described in detail below with reference to the accompanying drawings and in conjunction with the embodiments.

For ease of understanding, a conventional pixel interpolation scene is briefly introduced first with reference to FIG. 1 . This scene is a video encoding scene. The so-called video encoding method refers to the method of converting the original video format file into another video format file through compression technology.

As shown in FIG. 1A , the traditional video coding process usually includes processes such as prediction, transformation, quantization, and entropy coding, and finally outputs a video code stream. During prediction, pixel interpolation is usually required. The video encoding process is described below.

Video data has strong correlation, so there is a lot of redundant information. The redundant information can be divided into spatial domain redundant information and time domain redundant information. The purpose of prediction is to use the predicted image block data to reduce redundant information of the image block currently to be encoded, that is, to remove the correlation between the data. In the prediction stage, a predicted image block relative to the current image block to be encoded can be obtained. A residual block corresponding to the current image block to be encoded can be obtained by subtracting the pixel values of the current image block to be encoded and the predicted image block, thereby completing the prediction process.

Prediction usually includes two types of intra prediction and inter prediction. The intra-frame prediction is to predict the current image block to be encoded by using the image blocks already encoded in the same video frame, so as to reduce spatial redundancy information. Inter-frame prediction is to predict the current to-be-encoded image block by using the image blocks that have been encoded in the previous frame or multiple previous frames, so as to reduce redundant information in the temporal domain. Specifically, the principle of inter-frame prediction is to use the temporal correlation between adjacent frames in the video, use the previously encoded reconstructed frame as a reference frame, and predict the current to-be-encoded image block through motion estimation and motion compensation to remove the The temporal redundancy information in the video, and then achieve the purpose of compressing the video data. Among them, motion estimation refers to the need to find the image block most similar to the current image to be encoded in the reference frame, so that the encoding cost is the smallest and the encoding efficiency is the highest. Motion compensation is a process of modifying the searched image blocks on the basis of motion estimation.

Taking the H.265 video compression standard as an example, for the current image block to be encoded, in order to achieve high-precision motion estimation, it can be implemented in two steps: Integer Motion Estimation (IME) and Fractional Motion Estimation (Fraction) Motion Estimation, FME). Integer-pixel motion estimation is used to perform a matching search with integer-pixel accuracy within a certain range of the reference frame, and obtain the most matching image block with integer-pixel accuracy. Due to the continuity of motion of natural objects, the motion of objects between two adjacent frames may not be exactly separated by an integer pixel unit. Therefore, in order to improve the accuracy of motion estimation, it is necessary to perform pixel-by-pixel estimation on the basis of whole-pixel motion estimation. Match search is performed on sub-pixel precision (eg 1/2 pixel precision, 1/4 pixel precision, etc.). Since pixel values of sub-pixel positions do not exist in digital video, these pixel values need to be calculated by interpolation.

As an example, FIG. 1B shows a schematic diagram of sub-pixel positions. As shown in FIG. 1B , for a certain integer pixel point A(0,0) in a 4×4 image block, there are 8 1/2 pixel positions around the point. Further, for each 1/2 pixel position, there are 8 1/4 pixel positions around the position. Since the pixel values of sub-pixels are more accurate, the matching search of sub-pixels greatly improves the accuracy of motion estimation compared with the accuracy of whole pixels.

After the prediction process is performed, processes such as transformation, quantization, and entropy coding need to be performed in sequence. The transform operation is used to transform the residual block obtained in the prediction process from the time domain to the frequency domain. The transformation method may include discrete cosine transform (discrete cosine transform, DCT), discrete wavelet transform (discrete wavelet transform, DWT), and the like. After transforming the residual block, a transform coefficient matrix can be obtained. The quantization operation is used to reduce the coefficient of variation obtained after transformation, thereby facilitating entropy coding. Entropy coding, that is, coding in which no information is lost according to the principle of entropy during the coding process. Common entropy codes include: Shannon coding, Huffman coding and arithmetic coding (arithmetic coding). After the entropy coding process, an entropy coded bit stream can be obtained. The video encoding process can be completed by storing or sending the entropy encoded bit stream and related information (such as mode information, motion vector, etc.) to the decoding end.

It should be noted that, after quantization and transformation, a residual block (which may be called a reconstructed residual block) can also be recovered through inverse quantization and inverse transformation, and then the reconstructed residual block and the predicted image block are added to obtain the reconstruction. image block, and finally perform in-loop filtering on the reconstructed image block to obtain the final reconstructed image, which is then provided to the subsequent coded image for inter-frame prediction. The in-loop filtering may include, but is not limited to, deblocking filtering, pixel adaptive offset (Sample adaptive offset, SAO), and the like. Through the above process, the encoding of the video can be completed.

Since pixel-by-pixel motion estimation is required in the inter prediction stage of the video coding process, pixel interpolation is required. In the prior art, interpolation parameters are usually calculated in real time during the pixel interpolation process, such as the offset of the pixel matrix required by the interpolation point compared to the input pixel matrix, filtering parameters, etc., so the interpolation efficiency is low. Therefore, it is urgent to find a method that can improve the interpolation efficiency.

Please refer to FIG. 2, which shows a flowchart of a pixel interpolation method 200 according to the present application. The executing subject of the pixel interpolation method can be various electronic devices. For example cell phones, tablets, laptops, desktops, servers, etc.

The flow of the pixel interpolation method includes the following steps:

Step 201: Select a target interpolation position of a target image block.

In this embodiment, when the execution body of the pixel interpolation method (such as the electronic device described above) needs to perform interpolation on a target image block in a certain image, the target interpolation position of the target image block can be selected first. The target pixel block is an image block to be subjected to pixel interpolation, for example, it may be an image block obtained by performing an integer pixel search in the inter-frame prediction stage of the video coding process. The size of the target image block can be preset as required, for example, it can be set to 4×4, 8×8, etc., which is not specifically limited here.

The target interpolation position may be a position to be interpolated between integer pixels. As an example, when interpolation needs to be performed with 1/2 interpolation precision, 8 interpolation positions may be included around each integer pixel in the target image block. When interpolation needs to be performed with 1/4 interpolation precision, 8 interpolation positions may be further included around each 1/2 pixel point in the target image block. Therefore, each integer pixel in the target image block may correspond to multiple interpolation positions. Each interpolation position of each integer pixel can be numbered according to the same rule, so that a same numbered interpolation position of each integer pixel of the target image block is used as the target interpolation position.

Taking the distribution diagram of the interpolation positions of the pixel interpolation method shown in FIG. 3 as an example, if the size of the target pixel block is 4×4, the target pixel block includes 16 whole pixels. When the highest interpolation precision is 1/4, each integer pixel in the target image block can contain 49 interpolation positions (including the integer pixel itself). The 49 interpolation positions of each integer pixel can be numbered respectively to obtain a total of 49 numbers from 0 to 48. The target interpolation position can be the interpolation position corresponding to any number. For example, it can be the interpolation position No. 8 of each integer pixel in the target image block. Since the target image block has a total of 16 integer pixels, there are 16 No. 8 interpolation positions in total, and each No. 8 interpolation position can be regarded as the target interpolation position of the target image block.

Step 202 , from a preset mapping table, query the target pixel matrix offset corresponding to the target interpolation position and the target filtering parameters corresponding to the target interpolation position.

In this embodiment, a preset mapping table may be stored in the above-mentioned execution body. The mapping table may indicate a mapping relationship between interpolation positions and pixel matrix offsets, and may indicate a mapping relationship between interpolation positions and filtering parameters used for interpolation calculation. Thus, the above-mentioned execution body can query the target pixel matrix offset corresponding to the target interpolation position and the target filter parameter corresponding to the target interpolation position from the mapping table.

Here, the pixel matrix offset is the offset of the pixel matrix required for the interpolation position relative to the input pixel matrix. In practice, when an image block needs to be interpolated, it is usually necessary to input a pixel matrix to provide pixel data required for the interpolation process. For example, if a 4×4 image block is interpolated, a 12×12 pixel matrix needs to be input, and the pixel matrix is the input pixel matrix. However, the pixel matrix actually required for different interpolation positions in an image block is usually different from the input pixel matrix. For example, for a specific interpolation position in the image block, the actual required pixel matrix size is 11× 11. Compared with the 12×12 pixel matrix, the 11×11 pixel matrix may have an offset in the horizontal direction and the vertical direction, and the offset is the pixel matrix offset.

Taking the interpolation positions shown in FIG. 3 as an example, each integer pixel includes 49 interpolation positions (eg, interpolation positions numbered 0 to 48). Since the pixel matrix required for different interpolation positions is usually different from the input pixel matrix, the pixel matrix offset corresponding to each interpolation position can be pre-calculated, so as to map the 49 interpolation positions to the pixel matrix offset Relationships are pre-recorded in the mapping table. The above executive body only needs to look up the table during the interpolation process to obtain the target pixel matrix offset corresponding to the target interpolation position. Since the interpolation process does not need to calculate the offset of the pixel matrix in real time, the efficiency of pixel interpolation can be improved.

Similarly, because the filtering parameters used for different interpolation positions in the interpolation process are usually different (for example, the filtering parameters applicable to 1/4 pixel points are [-1,4,-10,58,17,-5,1 ], the filtering parameters applicable to 1/2 pixel points are [-1,4,-11,40,40,-11,4,-1], and the filtering parameters applicable to 3/4 pixel points are [1,- 5,17,58,-10,4,-1]), so the filtering parameters corresponding to the 49 interpolation positions can be pre-determined, so that the mapping relationship between the 49 interpolation positions and the filtering parameters can be recorded in the mapping table middle. The above executive body only needs to look up the table during the interpolation process to obtain the target filtering parameters corresponding to the target interpolation position. Since the interpolation process does not need to filter and determine the filtering parameters in real time, the efficiency of pixel interpolation can be improved.

In some optional embodiments, the above-mentioned mapping table may include first mapping information of the interpolation position and the offset of the pixel matrix. The first mapping information can represent the mapping relationship between the interpolation position and the offset of the pixel matrix, and can be generated by the following steps:

The first step is to obtain the input pixel matrix corresponding to the specified size image block.

Wherein, the size of the specified size image block is the same as the size of the target image block. As an example, the specified size image block and the above-mentioned target image block are both 4×4 image blocks, and both contain 16 integer pixels. The input pixel matrix corresponding to the specified size image block can be a 12×12 pixel matrix.

The second step is to determine the pixel matrix offset of the pixel matrix required for each interpolation position in the image block of the specified size relative to the input pixel matrix.

Here, for each interpolation position in the image block of the specified size (that is, the interpolation position corresponding to each position number, such as the interpolation position No. 8 in the 4×4 image block without an integer pixel point), the interpolation value can be determined first. The position of the pixel matrix required for the position, and then the pixel matrix offset is determined based on the relative positional relationship between the position and the position of the input pixel matrix.

The third step is to generate first mapping information based on the pixel matrix offset corresponding to each interpolation position.

Here, the mapping relationship between each interpolation position and the offset of the pixel matrix may be summarized in order to obtain the first mapping information. As an example, a position number may be set for each interpolation position first (as shown in FIG. 3 , which will not be repeated here); then, based on the pixel matrix offset corresponding to each interpolation position, a relationship between the position number and the pixel matrix offset is established. mapping to obtain the first mapping information.

In some optional embodiments, the pixel matrix offset corresponding to each interpolation position may include a horizontal pixel matrix offset and a vertical pixel matrix offset. Therefore, referring to FIG. 4 , the first mapping information may include the mapping relationship between each interpolation position (referred to by the available position number) and the offset of the pixel matrix in the horizontal direction, and may also include each interpolation position (referred to by the available position number) ) and the mapping relationship of the pixel matrix offset in the vertical direction.

In some optional embodiments, the mapping table may include second mapping information of interpolation positions and filtering parameters. The second mapping information may represent the mapping relationship between the interpolation position and the filtering parameter. Wherein, the second mapping information can be generated by the following steps:

The first step is to determine the type of each interpolation position in an image block of a specified size.

In the second step, the filtering parameters corresponding to each interpolation position are determined based on the type of each interpolation position.

In some examples, the type of interpolation location may include, but is not limited to, at least one of the following: 1/4 pixel, 1/2 pixel, 3/4 pixel, and integer pixel. Different types of interpolation positions can correspond to different filter parameters. You can directly select the default filter parameters for each type of interpolation position, or you can adjust the default filter parameters.

The third step is to generate second mapping information based on the filtering parameters corresponding to each interpolation position.

Here, the mapping relationship between each interpolation position and the filtering parameter can be summarized in order to obtain the second mapping information. As an example, a position number may be set for each interpolation position first (as shown in FIG. 3 , which will not be repeated here); then, an index number may be set for each filter parameter. For example, the index number of the filtering parameter of the whole pixel is 0. The index number of the filtering parameter of the 1/4 pixel is 1. The index number of the filtering parameter of the 1/2 pixel is 2. and the index number of the filtering parameter of the 3/4 pixel point is 3. Finally, based on the filtering parameters corresponding to each interpolation position, a mapping between the position number and the index number can be established to obtain the second mapping information.

In some optional embodiments, the type of each interpolation position includes the type of each interpolation position in the horizontal direction and the type of each interpolation position in the vertical direction. At this time, in the above second step, the filtering parameters corresponding to each interpolation position are determined based on the type of each interpolation position, which can be divided into the following steps: based on the type of each interpolation position in the horizontal direction, determine each interpolation position in the horizontal direction. Corresponding filtering parameters; based on the type of each interpolation position in the vertical direction, determine the filtering parameters corresponding to each interpolation position in the vertical direction.

On this basis, referring to FIG. 5 , the second mapping information may include the mapping relationship between each interpolation position and the filter parameter (represented by an index number) corresponding to the horizontal direction, and the corresponding interpolation position in the vertical direction. The mapping relationship of the filtering parameters (represented by index numbers).

In some optional embodiments, the filtering parameters corresponding to different types of interpolation positions may be parameter vectors with the same dimension. For example, the type of interpolation location may include, but is not limited to, at least one of the following: 1/4 pixel point, 1/2 pixel point, 3/4 pixel point, and whole pixel point. The filtering parameters corresponding to different types of interpolation positions are all 8-dimensional parameter vectors, and the 8-dimensional parameter vectors corresponding to different types of interpolation positions are different.

In some optional embodiments, the filtering parameters corresponding to the 1/4 pixel points are determined by the following steps: first, the first default filtering parameters corresponding to the 1/4 pixel points are obtained, and the first default filtering parameters are M-dimensional parameters vector, where M is a positive integer. After that, 0 is added to the end of the first default filter parameter to obtain the final filter parameter corresponding to the 1/4 pixel point, and the final filter parameter is an N-dimensional parameter vector, N=M+1. As an example, the first default filter parameter may be [-1, 4, -10, 58, 17, -5, 1], and M is 7. After adding 0 to the end of the first default filter parameter, the final filter parameter [-1,4,-10,58,17,-5,1,0] corresponding to the 1/4 pixel point can be obtained, at this time N is 8.

In some optional embodiments, the filter parameter corresponding to the 1/2 pixel point is the second default filter parameter, the second default filter parameter is an M-dimensional parameter vector, and M is a positive integer. For example, the second default filter parameter is [-1,4,-11,40,40,-11,4,-1], and M is 7 at this time.

In some optional embodiments, the filtering parameters corresponding to the 3/4 pixel points may be determined by the following steps: first, obtain the third default filtering parameters corresponding to the 3/4 pixel points, and the third default filtering parameters are M-dimensional parameter vector, M is a positive integer. After that, add 0 to the first position of the third default filter parameter to obtain the final filter parameter corresponding to the 3/4 pixel point, and the final filter parameter is an N-dimensional parameter vector, N=M+1. As an example, the third default filter parameter may be [1,-5,17,58,-10,4,-1], and M is 7. After adding 0 to the first position of the third default filter parameter, the final filter parameter [0,1,-5,17,58,-10,4,-1] corresponding to the 3/4 pixel point can be obtained. N is 8.

In some optional embodiments, the filtering parameter corresponding to an integer pixel is an M-dimensional parameter vector with a parameter value of 0, where M is a positive integer. That is, the filter parameter corresponding to the integer pixel is [0,0,0,0,0,0,0,0].

By using the filtering parameters corresponding to different types of interpolation positions as parameter vectors with the same dimension, it is convenient to use the same set of hardware architecture to interpolate different types of interpolation positions, thereby reducing hardware costs.

In some optional embodiments, among the filtering parameters corresponding to 1/4 pixel points, the filtering parameters corresponding to 1/2 pixel points, the filtering parameters corresponding to 3/4 pixel points, and the filtering parameters corresponding to integer pixels The parameter values of the same dimension are not less than 0 or are not greater than 0. As an example, the filtering parameters corresponding to 1/4 pixel points are [-1,4,-10,58,17,-5,1,0], and the filtering parameters corresponding to 1/2 pixel points are [-1, 4,-11,40,40,-11,4,-1], the filtering parameters corresponding to 3/4 pixel points are [0,1,-5,17,58,-10,4,-1], The filter parameter corresponding to an integer pixel is [0, 0, 0, 0, 0, 0, 0, 0], which ensures that the parameter values of the same dimension are not less than 0 or are not greater than 0.

By making the parameter values of the same dimension in the filtering parameters corresponding to different types of interpolation positions not less than 0 or not greater than 0, it is convenient to use the same set of hardware architecture (for example, including the same set of adders and subtractors) for different types of interpolation Positions are interpolated, thereby reducing hardware costs.

Step 203: Interpolate the target interpolation position based on the target pixel matrix offset and the target filter parameter.

In this embodiment, the execution subject may perform interpolation on the target interpolation position based on the target pixel matrix offset and the target filter parameter.

In some optional embodiments, the input pixel matrix corresponding to the target image block may be obtained first; then, the target pixel matrix required for interpolation of the target interpolation position may be determined based on the target pixel matrix offset and the input pixel matrix. Finally, the target interpolation position can be interpolated based on the pixel data in the target pixel matrix and the target filter parameters.

In some optional embodiments, when using the target pixel matrix and the target filtering parameters to interpolate the target interpolation position, the above-mentioned executing subject may first select reference pixels in the target pixel matrix based on the target interpolation position. Then, use the target filtering parameters to filter the reference pixels to obtain the interpolation result of the target interpolation position. In some examples, if the type of the target interpolation position is a 1/2 pixel point, an integer pixel point in the target pixel matrix may be selected as a reference pixel point. In other examples, if the type of the target interpolation position is a 1/4 pixel point or a 3/4 pixel point, a 1/2 pixel point in the target pixel matrix may be selected as a reference pixel point.

Taking the interpolation of 1/2 precision for the target image block of 4×4 size as an example, the position of interpolation required for 1/2 interpolation precision includes the upper left interpolation position of the whole pixel (as shown in Figure 3, the position number corresponding to No. 8) position), the left middle interpolation position (the position corresponding to the position number 22 in Figure 3), the lower left interpolation position (the position corresponding to the position number 36 in Figure 3), the upper right interpolation position (the position corresponding to the position number 12 in Figure 3) number corresponding to the position number), the right middle interpolation position (the position corresponding to the 26th position number in Figure 3), the lower right interpolation position (the position corresponding to the 40th position number in Figure 3), the upper middle interpolation position (the position corresponding to the position number 10 in Figure 3), the middle interpolation position (the position corresponding to the position number 24 in Figure 3, that is, the position of the whole pixel), and the middle and lower interpolation positions (as shown in Figure 3) The position corresponding to the position number 38).

If the current target interpolation position is the left middle interpolation position (that is, the position corresponding to the 22nd position number) or the right middle interpolation position (that is, the position corresponding to the 26th position number) of each integer pixel, the target pixel matrix can be used. The pixels in the horizontal direction (including the 4 integer pixels on the left side of the target interpolation position and the 4 integer pixels on the right side) are used as reference pixels, and the target filtering parameters corresponding to the target interpolation position are used for the reference pixels. The pixel value is filtered and calculated to obtain the pixel value of the target interpolation position.

If the current target interpolation position is the upper-middle interpolation position (that is, the position corresponding to the position number 10) or the lower-middle interpolation position (that is, the position corresponding to the position number 38) of each integer pixel point, you can use the target pixel matrix. The pixels in the vertical direction (including the 4 integer pixels above the target interpolation position and the 4 integer pixels below) are used as reference pixels, and the target filtering parameters corresponding to the target interpolation position are used. The value is filtered and calculated to obtain the pixel value of the target interpolation position.

If the current target interpolation position is the upper left interpolation position of each integer pixel (that is, the position corresponding to the position number 8) or the lower left interpolation position (that is, the position corresponding to the position number 36), you can use the middle left of each integer pixel. The pixel at the interpolation position is used as a reference pixel point, and the pixel value of the reference pixel point is filtered and calculated by using the target filter parameter corresponding to the target interpolation position to obtain the pixel value of the target interpolation position.

If the current target interpolation position is the upper right interpolation position of each integer pixel (that is, the position corresponding to the position number 12) or the lower right interpolation position (that is, the position corresponding to the position number 40), the right interpolation position of each integer pixel can be used. The pixel at the intermediate interpolation position is used as a reference pixel point, and the pixel value of the reference pixel point is filtered and calculated by using the target filtering parameter corresponding to the target interpolation position to obtain the pixel value of the target interpolation position.

If the current target interpolation position is the middle interpolation position of each integer pixel (that is, the position corresponding to the position number 24), since the position of the integer pixel does not require interpolation, the pixel value of the integer pixel can be directly used as the target interpolation position. Pixel values.

In practical applications, the interpolation position in the horizontal direction of each integer pixel in the target image block may be firstly used as the target interpolation position for interpolation. Specifically, the interpolation can be performed first by using the middle left interpolation position of each pixel as the target interpolation position, and then using the middle right interpolation position of each integer pixel in the target image block as the target interpolation position for interpolation, and then using the target image block in the interpolation position. The median interpolation position of each integer pixel of , is used as the target interpolation position for interpolation, so as to complete the interpolation of the interpolation position in the horizontal direction.

Among them, the interpolation of the left-middle interpolation position can be divided into two steps. The first time is to interpolate the left middle position of the left two columns of integer pixels, and the second time is to interpolate the left middle position of the right two columns of integer pixels. Similarly, when interpolating the right-middle interpolation position, it can also be divided into two steps. The first time is to interpolate the right middle position of the left two columns of integer pixels, and the second time is to interpolate the right middle position of the right two columns of integer pixels. The interpolation result of each time can be stored in the cache, and in the next interpolation, for the repeated interpolation position, the previous interpolation result can be directly extracted from the buffer without repeated calculation. When interpolating the median interpolation position, since the median interpolation position is the position of the integer pixel, the pixel value of the integer pixel can be directly used.

After the interpolation positions of each integer pixel in the horizontal direction are used as the target interpolation positions, the interpolation positions in the vertical direction may be used as the target interpolation positions for interpolation. Specifically, the upper left interpolation position, the middle left interpolation position (the obtained interpolation result can be directly read), and the lower left interpolation position of each integer pixel can be interpolated first as the target interpolation position; then, the interpolation position of each integer pixel can be The upper right interpolation position, the middle right interpolation position (the obtained interpolation result can be directly read), and the lower right interpolation position are used as the target interpolation position for interpolation; Directly read the pixel value of the whole pixel), the middle and lower interpolation positions are used as the target interpolation positions for interpolation. In this way, all the interpolation processes under the full 1/2 interpolation precision can be completed.

Wherein, when performing interpolation on each position to be interpolated in the vertical direction, the interpolation may also be performed twice, which will not be repeated here.

It should be noted that under the 1/4 interpolation precision, the interpolation result under the 1/2 interpolation precision can be used for calculation, and the calculation process is basically the same as the interpolation process under the 1/2 interpolation precision, and will not be repeated here. In addition, the interpolation order in practical application is not limited to the above example, and other interpolation orders can also be set as required. In addition, the pixel interpolation method of the embodiment of the present application is not limited to the 1/4 interpolation precision and the 1/2 interpolation precision, and can also be applied to other interpolation precisions, which will not be repeated here.

In some optional embodiments, the target filtering parameter is an N-dimensional parameter vector, where N is a positive integer (eg, 8). The above executive body can perform interpolation on the target interpolation position with the help of the same hardware framework. Specifically, the above-mentioned execution body controls the hardware framework to calculate the interpolation result of the target interpolation position according to the following steps:

The first step is to control N data selectors to select N data, that is, each data selector selects one data. Among them, the N pieces of data are respectively the data processing results after processing the target pixel matrix by using the N parameter values in the target filtering parameters.

A schematic diagram of the hardware framework used to calculate the interpolation results can be found in Figure 6. As shown in Figure 6, the hardware frame can contain 8 data selectors, numbered 0 to 7 in sequence. If the target filter parameter is [-1,4,-11,40,40,-11,4,-1], the data selector numbered 0 can select the first parameter (ie -1) to the target pixel matrix The data processing result after processing; the data selector numbered 1 can select the second parameter (ie, 4) the data processing result after processing the target pixel matrix; and so on.

Here, the N data selectors may be pre-divided into the first group and the second group. Taking the 8 data selectors shown in Figure 3 as an example, the data selectors numbered 0, 2, 5, and 7 can be divided into the first group, and the data selectors numbered 1, 3, 4, and 6 can be divided into the first group. Divide into the second group.

In the second step, the data output by the data selectors in the first group is input into the first adder, and the data output by the data selectors in the second group is input into the second adder.

As an example, referring to FIG. 3, in the first group of data selectors, the data output by the data selectors numbered 0 and 7 can be added by an adder and then input to the first adder, and the data numbered 2 and 5 are selected The data output by the adder can be added by another adder and then input to the first adder. In the second group of data selectors, the data output by the data selectors numbered 1 and 6 can be added by one adder and then input to the second adder, and the data output by the data selectors numbered 3 and 4 can be added by another adder. One adder adds and inputs to the second adder. Thus, the effect of inputting the data outputted by the data selectors in the first group to the first adder and inputting the data outputted by the data selectors in the second group to the second adder can be achieved.

In the third step, the data output by the first adder and the data output by the second adder are input to the subtractor, and the data output by the subtracter is used as the interpolation result of the target interpolation position. The subtractor may subtract the data output from the second adder and the data output from the first adder to obtain a difference result.

Here, the filtering parameters corresponding to different types of interpolation positions can be set as N-dimensional (eg, 8-dimensional) parameter vectors, and the parameter values of the same dimension in the filtering parameters are all not less than 0 or not greater than 0. Through the above settings, the same set of hardware frameworks can be used to calculate the interpolation results of different interpolation precisions and different types of interpolation positions, and there is no need to set different hardware frameworks for different interpolation precisions and different types of interpolation positions, thereby saving hardware resources.

With the pixel interpolation method provided by the above embodiments of the present application, the target interpolation position of the target image block is selected, and then the target pixel matrix offset corresponding to the target interpolation position and the target pixel matrix offset corresponding to the target interpolation position are queried from the preset mapping table. target filter parameters, so that the target interpolation position can be interpolated based on the target pixel matrix offset and the target filter parameters. Therefore, in the interpolation process, it is only necessary to look up the table to obtain the target pixel matrix offset corresponding to the target interpolation position and the target filtering parameters. The interpolation process does not need to calculate the pixel matrix offset and determine the filtering parameters in real time. Therefore, Improves the efficiency of pixel interpolation.

Further, the filtering parameters corresponding to different types of interpolation positions are set as parameter vectors of the same dimension, and the parameter values of the same dimension in the filtering parameters are not less than 0 or are not greater than 0, and the same set of hardware frameworks can be used to calculate different values. Interpolation precision and interpolation results for different types of interpolation positions. Thereby, hardware resources are saved.

Further referring to FIG. 7 , as an implementation of the methods shown in the above figures, the present application provides an embodiment of a pixel interpolation apparatus, and the apparatus embodiment corresponds to the method embodiment shown in FIG. 5 . The electronic device can be among various electronic devices.

As shown in FIG. 7 , the above-mentioned pixel interpolation 700 in this embodiment includes: a selection unit 701 configured to select a target interpolation position of a target image block; a query unit 702 configured to query the above target from a preset mapping table The target pixel matrix offset corresponding to the interpolation position and the target filtering parameter corresponding to the above-mentioned target interpolation position; the interpolation unit 703 is configured to perform interpolation on the above-mentioned target interpolation position based on the above-mentioned target pixel matrix offset and the above-mentioned target filtering parameter.

In some optional implementations of this embodiment, the above-mentioned mapping table includes first mapping information of the interpolation position and the offset of the pixel matrix; the above-mentioned first mapping information is generated by the following steps: obtaining the input corresponding to the image block of the specified size A pixel matrix, wherein the size of the image block of the specified size is the same as that of the target image block; the pixel matrix offset required for each interpolation position in the image block of the specified size is determined relative to the pixel matrix of the input pixel matrix; based on the above The pixel matrix offset corresponding to each interpolation position is used to generate the first mapping information.

In some optional implementations of this embodiment, generating the first mapping information based on the pixel matrix offsets corresponding to the above-mentioned interpolation positions includes: setting a position number for each of the above-mentioned interpolation positions; The offset of the pixel matrix is established, and the mapping between the position number and the offset of the pixel matrix is established to obtain the first mapping information.

In some optional implementations of this embodiment, the pixel matrix offset corresponding to each interpolation position includes a horizontal pixel matrix offset and a vertical pixel matrix offset; the first mapping information includes The mapping relationship between the above-mentioned interpolation positions and the offset of the pixel matrix in the horizontal direction, and the mapping relationship between the above-mentioned interpolation positions and the offset of the pixel matrix in the vertical direction.

In some optional implementations of this embodiment, the above-mentioned mapping table includes second mapping information of interpolation positions and filtering parameters; the above-mentioned second mapping information is generated by the following steps: determining the value of each interpolation position in an image block of a specified size type, wherein the size of the image block of the specified size is the same as the size of the target image block; based on the type of each interpolation position, the filtering parameters corresponding to the interpolation positions are determined; based on the filtering parameters corresponding to the interpolation positions, a second mapping is generated information.

In some optional implementations of this embodiment, generating the second mapping information based on the filtering parameters corresponding to the above-mentioned interpolation positions includes: setting a position number for each of the above-mentioned interpolation positions; setting an index number for each filtering parameter; For the filtering parameters corresponding to the above-mentioned interpolation positions, a mapping between the position number and the index number is established to obtain the second mapping information.

In some optional implementations of this embodiment, the types of the above-mentioned interpolation positions include the types of the above-mentioned interpolation positions in the horizontal direction and the types of the above-mentioned interpolation positions in the vertical direction.

In some optional implementations of this embodiment, determining the filtering parameters corresponding to the interpolation positions based on the types of the interpolation positions includes: determining the horizontal direction based on the types of the interpolation positions in the horizontal direction. filtering parameters corresponding to the above-mentioned interpolation positions; based on the types of the above-mentioned interpolation positions in the vertical direction, the filtering parameters corresponding to the above-mentioned interpolation positions in the vertical direction are determined.

In some optional implementations of this embodiment, the second mapping information includes a mapping relationship between the interpolation positions and the filtering parameters corresponding to the horizontal direction, and includes the interpolation positions corresponding to the vertical direction. The mapping relationship of the filtering parameters.

In some optional implementations of this embodiment, the type of the interpolation position includes at least one of the following: 1/4 pixel point, 1/2 pixel point, 3/4 pixel point, and whole pixel point.

In some optional implementations of this embodiment, the filtering parameters corresponding to different types of interpolation positions are parameter vectors with the same dimension.

In some optional implementation manners of this embodiment, the filtering parameters corresponding to the above-mentioned 1/4 pixel points are determined by the following steps: obtaining the first default filtering parameters corresponding to the above-mentioned 1/4 pixel points, the above-mentioned first default filtering parameters The parameter is an M-dimensional parameter vector, and M is a positive integer; add 0 to the end of the first default filtering parameter above to obtain the final filtering parameter corresponding to the above-mentioned 1/4 pixel point, and the above-mentioned final filtering parameter is an N-dimensional parameter vector, N=M+1.

In some optional implementations of this embodiment, the filtering parameters corresponding to the above-mentioned 1/2 pixel points are second default filtering parameters, the above-mentioned second default filtering parameters are M-dimensional parameter vectors, and M is a positive integer.

In some optional implementations of this embodiment, the filtering parameters corresponding to the above-mentioned 3/4 pixel points are determined by the following steps: obtaining the third default filtering parameters corresponding to the above-mentioned 3/4 pixel points, and the above-mentioned third default filtering parameters The parameter is an M-dimensional parameter vector, and M is a positive integer; add 0 to the first position of the above-mentioned third default filtering parameter to obtain the final filtering parameter corresponding to the above-mentioned 3/4 pixel point, and the above-mentioned final filtering parameter is an N-dimensional parameter vector. , N=M+1.

In some optional implementations of this embodiment, the filtering parameter corresponding to the above-mentioned integer pixel is an M-dimensional parameter vector with a parameter value of 0, where M is a positive integer.

In some optional implementation manners of this embodiment, the filtering parameters corresponding to the foregoing 1/4 pixel points, the filtering parameters corresponding to the foregoing 1/2 pixel points, the filtering parameters corresponding to the foregoing 3/4 pixel points, and the above The parameter values of the same dimension in the filtering parameters corresponding to the integer pixels are not less than 0 or are not greater than 0.

In some optional implementations of this embodiment, the above-mentioned interpolation unit 703 is further configured to: obtain the input pixel matrix corresponding to the above-mentioned target image block; The target pixel matrix required for interpolation of the target interpolation position; based on the target pixel matrix and the target filtering parameter, the target interpolation position is interpolated.

In some optional implementations of this embodiment, the above-mentioned interpolation unit 703 is further configured to: select a reference pixel point in the above-mentioned target pixel matrix based on the type of the above-mentioned target interpolation position; The pixel points are filtered to obtain the interpolation result of the target interpolation position.

In some optional implementations of this embodiment, the above-mentioned interpolation unit 703 is further configured to: if the type of the above-mentioned target interpolation position is a 1/2 pixel point, select an integer pixel point in the above-mentioned target pixel matrix, As a reference pixel; if the type of the target interpolation position is a 1/4 pixel or a 3/4 pixel, then select a 1/2 pixel in the target pixel matrix as a reference pixel.

In some optional implementations of this embodiment, the above-mentioned target filtering parameter is an N-dimensional parameter vector, and N is a positive integer; and, the above-mentioned interpolation unit 703 is further configured to: control N data selectors to select N pieces of data , the above-mentioned N data are respectively the data processing result after the above-mentioned target pixel matrix is processed by using the N parameter values in the target filtering parameters, wherein, the above-mentioned N data selectors are pre-divided into the first group and the second group; Input the data outputted by the data selectors in the above-mentioned first group into the first adder, and input the data outputted by the data selectors in the above-mentioned second group into the second adder; The data output by the second adder is input to the subtracter, and the data output by the subtracter is used as the interpolation result of the target interpolation position.

The pixel interpolation device provided by the above embodiments of the present application selects the target interpolation position of the target image block, and then queries the target pixel matrix offset corresponding to the target interpolation position and the target corresponding to the target interpolation position from the preset mapping table. filter parameters, so that the target interpolation position can be interpolated based on the target pixel matrix offset and the target filter parameter. Therefore, in the interpolation process, it is only necessary to look up the table to obtain the target pixel matrix offset corresponding to the target interpolation position and the target filtering parameters. The interpolation process does not need to calculate the pixel matrix offset and determine the filtering parameters in real time. Therefore, Improves the efficiency of pixel interpolation.

As an implementation of the above method shown in FIG. 1 , the present application provides an embodiment of an electronic device, which corresponds to the method embodiment shown in FIG. 1 . Referring to FIG. 8 , the electronic device may specifically include: a processor 801 and a memory 802 .

The above-mentioned memory 801 can be used to store program instructions.

The above-mentioned processor 802 can be used to execute the program instructions stored in the above-mentioned memory. When the program instructions are executed, the above-mentioned processor can be used to perform the following steps: selecting the target interpolation position of the target image block; from the preset mapping table, The target pixel matrix offset corresponding to the target interpolation position and the target filtering parameter corresponding to the target interpolation position are queried; based on the target pixel matrix offset and the target filtering parameter, the target interpolation position is interpolated.

In some optional implementations of this embodiment, the above-mentioned processor 802 is further configured to: obtain an input pixel matrix corresponding to the above-mentioned target image block; The target pixel matrix required for target interpolation position interpolation; based on the target pixel matrix and the target filtering parameter, the target interpolation position is interpolated.

In some optional implementations of this embodiment, the above-mentioned processor 802 is further configured to: select reference pixels in the above-mentioned target pixel matrix based on the type of the above-mentioned target interpolation position; Filter the points to obtain the interpolation result of the above target interpolation position.

In some optional implementations of this embodiment, the above-mentioned processor 802 is further configured to: if the type of the above-mentioned target interpolation position is a 1/2 pixel point, select an integer pixel point in the above-mentioned target pixel matrix as the type of the above-mentioned target pixel matrix. Reference pixel; if the type of the target interpolation position is 1/4 pixel or 3/4 pixel, then select 1/2 pixel in the target pixel matrix as the reference pixel.

In some optional implementations of this embodiment, the target filtering parameter is an N-dimensional parameter vector, and N is a positive integer; and the processor 802 is further configured to: control the N data selectors to select N data, The above-mentioned N pieces of data are respectively the data processing results after the above-mentioned target pixel matrix is processed by using the N parameter values in the target filtering parameters, wherein the above-mentioned N data selectors are pre-divided into the first group and the second group; The data output by the data selectors in the first group is input to the first adder, and the data output by the data selectors in the second group is input into the second adder; the data output by the first adder and the above The data output by the second adder is input to the subtractor, and the data output by the subtractor is used as the interpolation result of the target interpolation position.

The electronic device provided by the above embodiments of the present application selects the target interpolation position of the target image block, and then queries the target pixel matrix offset corresponding to the target interpolation position and the target corresponding to the target interpolation position from the preset mapping table. filter parameters, so that the target interpolation position can be interpolated based on the target pixel matrix offset and the target filter parameter. Therefore, in the interpolation process, it is only necessary to look up the table to obtain the target pixel matrix offset corresponding to the target interpolation position and the target filtering parameters. The interpolation process does not need to calculate the pixel matrix offset and determine the filtering parameters in real time. Therefore, Improves the efficiency of pixel interpolation.

As for the electronic device embodiment, since it is basically similar to the method embodiment, the description is relatively simple, and for related parts, please refer to the partial description of the method embodiment.

Embodiments of the present application further provide a computer-readable medium, where a computer program is stored on the computer-readable medium. When the computer program is executed by a processor, each process of the above-mentioned embodiments of the method can be achieved, and the same technical effect can be achieved. In order to avoid repetition, when the computer program is executed by the processor, each process of the embodiments of the above-mentioned methods is implemented, which will not be repeated here.

The various embodiments in this specification are described in a progressive manner, and each embodiment focuses on the differences from other embodiments, and the same and similar parts between the various embodiments may be referred to each other.

It should be understood by those skilled in the art that the embodiments of the present application may be provided as a method, an apparatus, or a computer program product. Accordingly, the present application may take the form of an entirely hardware embodiment, an entirely software embodiment, or an embodiment combining software and hardware aspects. Furthermore, the present application may take the form of a computer program product embodied on one or more computer-readable media having computer-usable program code embodied therein, including but not limited to disk storage, CD-ROM, optical storage, and the like.

The present application is described with reference to flowchart illustrations and/or block diagrams of methods, terminal devices (systems), and computer program products according to the present application. It will be understood that each flow and/or block in the flowchart illustrations and/or block diagrams, and combinations of flows and/or blocks in the flowchart illustrations and/or block diagrams, can be implemented by computer program instructions. These computer program instructions may be provided to the processor of a general purpose computer, special purpose computer, embedded processor or other programmable data processing terminal equipment to produce a machine that causes the instructions to be executed by the processor of the computer or other programmable data processing terminal equipment Means are created for implementing the functions specified in the flow or flows of the flowcharts and/or the blocks or blocks of the block diagrams.

These computer program instructions may also be stored in a computer readable memory capable of directing a computer or other programmable data processing terminal equipment to operate in a particular manner, such that the instructions stored in the computer readable memory result in an article of manufacture comprising instruction means, the The instruction means implement the functions specified in the flow or flow of the flowcharts and/or the block or blocks of the block diagrams.

These computer program instructions can also be loaded on a computer or other programmable data processing terminal equipment, so that a series of operational steps are performed on the computer or other programmable terminal equipment to produce a computer-implemented process, thereby executing on the computer or other programmable terminal equipment The instructions executed on the above provide steps for implementing the functions specified in the flowchart or blocks and/or the block or blocks of the block diagrams.

While the preferred embodiments of the present application have been described, additional changes and modifications to these embodiments may occur to those skilled in the art once the basic inventive concepts are known. Therefore, the appended claims are intended to be construed to include the preferred embodiment and all changes and modifications that fall within the scope of this application.

Finally, it should also be noted that in this document, relational terms such as first and second are used only to distinguish one entity or operation from another, and do not necessarily require or imply these entities or that there is any such actual relationship or sequence between operations. Moreover, the terms "comprising", "comprising" or any other variation thereof are intended to encompass non-exclusive inclusion such that a process, method, article or terminal device that includes a list of elements includes not only those elements, but also a non-exclusive list of elements. other elements, or also include elements inherent to such a process, method, article or terminal equipment. Without further limitation, an element defined by the phrase "comprises a..." does not preclude the presence of additional identical elements in the process, method, article or terminal device comprising said element.

The methods, devices, electronic devices and computer-readable media provided by the present application have been described in detail above. The principles and implementations of the present application are described with specific examples. The descriptions of the above embodiments are only used for Help to understand the method of the present application and its core idea; meanwhile, for those of ordinary skill in the art, according to the idea of the present application, there will be changes in the specific implementation and application scope. In summary, the content of this specification It should not be construed as a limitation of this application.

Claims

A pixel interpolation method, characterized in that the method comprises:

Select the target interpolation position of the target image block;

From the preset mapping table, query the target pixel matrix offset corresponding to the target interpolation position and the target filtering parameter corresponding to the target interpolation position;

The target interpolation position is interpolated based on the target pixel matrix offset and the target filter parameter.
The method according to claim 1, wherein the mapping table includes first mapping information of the interpolation position and the offset of the pixel matrix;

The first mapping information is generated by the following steps:

obtaining an input pixel matrix corresponding to an image block of a specified size, wherein the size of the image block of the specified size is the same as that of the target image block;

determining the pixel matrix offset of the pixel matrix relative to the input pixel matrix required for each interpolation position in the image block of the specified size;

First mapping information is generated based on the pixel matrix offset corresponding to each interpolation position.
The method according to claim 2, wherein the generating the first mapping information based on the pixel matrix offset corresponding to each interpolation position comprises:

setting a position number for each of the interpolation positions;

Based on the pixel matrix offset corresponding to each interpolation position, a mapping between the position number and the pixel matrix offset is established to obtain first mapping information.
The method according to claim 2, wherein the pixel matrix offset corresponding to each interpolation position comprises a horizontal pixel matrix offset and a vertical pixel matrix offset;

The first mapping information includes the mapping relationship between the interpolation positions and the offset of the pixel matrix in the horizontal direction, and includes the mapping relationship between the interpolation positions and the offset of the pixel matrix in the vertical direction.
The method according to claim 1, wherein the mapping table includes second mapping information of interpolation positions and filtering parameters;

The second mapping information is generated through the following steps:

determining the type of each interpolation position in an image block of a specified size, wherein the size of the image block of the specified size is the same as the size of the target image block;

Based on the type of each interpolation position, determine the filtering parameter corresponding to each interpolation position;

The second mapping information is generated based on the filtering parameters corresponding to the interpolation positions.
The method according to claim 5, wherein the generating the second mapping information based on the filtering parameters corresponding to the interpolation positions comprises:

setting a position number for each of the interpolation positions;

Set the index number for each filter parameter;

Based on the filtering parameters corresponding to the interpolation positions, a mapping between position numbers and index numbers is established to obtain second mapping information.
The method according to claim 5, wherein the types of the interpolation positions include the types of the interpolation positions in the horizontal direction and the types of the interpolation positions in the vertical direction.
The method according to claim 7, wherein the determining the filtering parameters corresponding to the interpolation positions based on the types of the interpolation positions comprises:

determining the filtering parameters corresponding to the interpolation positions in the horizontal direction based on the types of the interpolation positions in the horizontal direction;

Based on the types of the interpolation positions in the vertical direction, filter parameters corresponding to the interpolation positions in the vertical direction are determined.
The method according to claim 7, wherein the second mapping information includes a mapping relationship between the interpolation positions and the filtering parameters corresponding to the horizontal direction, and includes the interpolation positions and the vertical The mapping relationship of the corresponding filter parameters in the direction.
The method according to claim 5, wherein the type of the interpolation position includes at least one of the following: 1/4 pixel point, 1/2 pixel point, 3/4 pixel point and whole pixel point.
The method according to claim 5, wherein the filtering parameters corresponding to different types of interpolation positions are parameter vectors with the same dimension.
The method according to claim 10, wherein the filtering parameters corresponding to the 1/4 pixel points are determined by the following steps:

Obtain the first default filter parameter corresponding to the 1/4 pixel point, the first default filter parameter is an M-dimensional parameter vector, and M is a positive integer;

Add 0 to the end of the first default filter parameter to obtain the final filter parameter corresponding to the 1/4 pixel point, where the final filter parameter is an N-dimensional parameter vector, N=M+1.
The method according to claim 10, wherein the filter parameter corresponding to the 1/2 pixel point is a second default filter parameter, the second default filter parameter is an M-dimensional parameter vector, and M is a positive integer.
The method according to claim 10, wherein the filtering parameters corresponding to the 3/4 pixel points are determined by the following steps:

Obtain the third default filter parameter corresponding to the 3/4 pixel point, the third default filter parameter is an M-dimensional parameter vector, and M is a positive integer;

0 is added before the first digit of the third default filtering parameter to obtain the final filtering parameter corresponding to the 3/4 pixel point, and the final filtering parameter is an N-dimensional parameter vector, N=M+1.
The method according to claim 10, wherein the filtering parameter corresponding to the integer pixel is an M-dimensional parameter vector with a parameter value of 0, and M is a positive integer.
The method according to claim 10, wherein the filtering parameters corresponding to the 1/4 pixel points, the filtering parameters corresponding to the 1/2 pixel points, and the filtering parameters corresponding to the 3/4 pixel points The parameter value and the parameter value of the same dimension in the filtering parameter corresponding to the integer pixel are not less than 0 or are not greater than 0.
The method according to claim 1, wherein the performing interpolation on the target interpolation position based on the target pixel matrix offset and the target filtering parameter comprises:

obtaining the input pixel matrix corresponding to the target image block;

Based on the target pixel matrix offset and the input pixel matrix, determine a target pixel matrix required for interpolation of the target interpolation position;

The target interpolation position is interpolated based on the target pixel matrix and the target filter parameter.
The method according to claim 17, wherein the performing interpolation on the target interpolation position based on the target pixel matrix and the target filtering parameter comprises:

Based on the type of the target interpolation position, select the reference pixel point in the target pixel matrix;

Filter processing is performed on the reference pixel point based on the target filtering parameter to obtain an interpolation result of the target interpolation position.
The method according to claim 18, wherein the selecting a reference pixel point in the target pixel matrix based on the type of the target interpolation position comprises:

If the type of the target interpolation position is a 1/2 pixel point, then select an integer pixel point in the target pixel matrix as a reference pixel point;

If the type of the target interpolation position is a 1/4 pixel point or a 3/4 pixel point, a 1/2 pixel point in the target pixel matrix is selected as a reference pixel point.
The method according to claim 18, wherein the target filtering parameter is an N-dimensional parameter vector, and N is a positive integer; and,

The filtering processing is performed on the reference pixel based on the target filtering parameter to obtain an interpolation result of the target interpolation position, including:

Controlling N data selectors to select N pieces of data, the N pieces of data are respectively the data processing results after the target pixel matrix is processed by using the N parameter values in the target filter parameters, wherein the N pieces of data are selected The device is pre-divided into the first group and the second group;

inputting the data outputted by the data selectors in the first group into the first adder, and inputting the data outputted by the data selectors in the second group into the second adder;

The data output by the first adder and the data output by the second adder are input to a subtractor, and the data output by the subtracter is used as the interpolation result of the target interpolation position.
A pixel interpolation device, characterized in that the device comprises:

a selection unit, configured to select the target interpolation position of the target image block;

a query unit, configured to query the target pixel matrix offset corresponding to the target interpolation position and the target filtering parameter corresponding to the target interpolation position from a preset mapping table;

An interpolation unit configured to interpolate the target interpolation position based on the target pixel matrix offset and the target filter parameter.
The device according to claim 21, wherein the mapping table includes first mapping information of the interpolation position and the offset of the pixel matrix;

The first mapping information is generated by the following steps:

obtaining an input pixel matrix corresponding to an image block of a specified size, wherein the size of the image block of the specified size is the same as that of the target image block;

determining the pixel matrix offset of the pixel matrix relative to the input pixel matrix required for each interpolation position in the image block of the specified size;

First mapping information is generated based on the pixel matrix offset corresponding to each interpolation position.
The device according to claim 22, wherein the generating the first mapping information based on the pixel matrix offset corresponding to each interpolation position comprises:

setting a position number for each of the interpolation positions;

Based on the pixel matrix offset corresponding to each interpolation position, a mapping between the position number and the pixel matrix offset is established to obtain the first mapping information.
The device according to claim 22, wherein the pixel matrix offset corresponding to each interpolation position comprises a pixel matrix offset in a horizontal direction and a pixel matrix offset in a vertical direction;

The first mapping information includes the mapping relationship between the interpolation positions and the offset of the pixel matrix in the horizontal direction, and includes the mapping relationship between the interpolation positions and the offset of the pixel matrix in the vertical direction.
The apparatus according to claim 21, wherein the mapping table includes second mapping information of interpolation positions and filtering parameters;

The second mapping information is generated through the following steps:

determining the type of each interpolation position in an image block of a specified size, wherein the size of the image block of the specified size is the same as the size of the target image block;

Based on the type of each interpolation position, determine the filtering parameter corresponding to each interpolation position;

The second mapping information is generated based on the filtering parameters corresponding to the interpolation positions.
The device according to claim 25, wherein the generating the second mapping information based on the filtering parameters corresponding to the interpolation positions comprises:

setting a position number for each of the interpolation positions;

Set the index number for each filter parameter;

Based on the filtering parameters corresponding to the interpolation positions, a mapping between position numbers and index numbers is established to obtain second mapping information.
The apparatus according to claim 25, wherein the types of the interpolation positions include the types of the interpolation positions in the horizontal direction and the types of the interpolation positions in the vertical direction.
The device according to claim 27, wherein the determining the filtering parameters corresponding to the interpolation positions based on the types of the interpolation positions comprises:

determining the filtering parameters corresponding to the interpolation positions in the horizontal direction based on the types of the interpolation positions in the horizontal direction;

Based on the types of the interpolation positions in the vertical direction, filter parameters corresponding to the interpolation positions in the vertical direction are determined.
The apparatus according to claim 27, wherein the second mapping information includes a mapping relationship between the interpolation positions and the filtering parameters corresponding to the horizontal direction, and includes the interpolation positions and the vertical The mapping relationship of the corresponding filter parameters in the direction.
The apparatus according to claim 25, wherein the type of interpolation position includes at least one of the following: 1/4 pixel point, 1/2 pixel point, 3/4 pixel point and whole pixel point.
The apparatus according to claim 25, wherein the filtering parameters corresponding to different types of interpolation positions are parameter vectors with the same dimension.
The device according to claim 30, wherein the filtering parameters corresponding to the 1/4 pixel points are determined by the following steps:

Obtain the first default filter parameter corresponding to the 1/4 pixel point, the first default filter parameter is an M-dimensional parameter vector, and M is a positive integer;

Add 0 to the end of the first default filtering parameter to obtain the final filtering parameter corresponding to the 1/4 pixel point, and the final filtering parameter is an N-dimensional parameter vector, N=M+1.
The apparatus according to claim 30, wherein the filtering parameter corresponding to the 1/2 pixel point is a second default filtering parameter, the second default filtering parameter is an M-dimensional parameter vector, and M is a positive integer.
The device according to claim 30, wherein the filtering parameters corresponding to the 3/4 pixel points are determined by the following steps:

Obtain the third default filter parameter corresponding to the 3/4 pixel point, the third default filter parameter is an M-dimensional parameter vector, and M is a positive integer;

0 is added before the first digit of the third default filtering parameter to obtain the final filtering parameter corresponding to the 3/4 pixel point, and the final filtering parameter is an N-dimensional parameter vector, N=M+1.
The device according to claim 30, wherein the filtering parameter corresponding to the integer pixel is an M-dimensional parameter vector with a parameter value of 0, and M is a positive integer.
The device according to claim 30, wherein the filtering parameters corresponding to the 1/4 pixel points, the filtering parameters corresponding to the 1/2 pixel points, and the filtering parameters corresponding to the 3/4 pixel points The parameter value and the parameter value of the same dimension in the filtering parameter corresponding to the integer pixel are not less than 0 or are not greater than 0.
The apparatus according to claim 21, wherein the interpolation unit is further configured to:

obtaining the input pixel matrix corresponding to the target image block;

Based on the offset of the target pixel matrix and the input pixel matrix, determine the target pixel matrix required for the interpolation of the target interpolation position;

The target interpolation position is interpolated based on the target pixel matrix and the target filter parameter.
The apparatus according to claim 27, wherein the interpolation unit is further configured to:

Based on the type of the target interpolation position, select the reference pixel point in the target pixel matrix;

Filter processing is performed on the reference pixel point based on the target filtering parameter to obtain an interpolation result of the target interpolation position.
The apparatus according to claim 28, wherein the interpolation unit is further configured to:

If the type of the target interpolation position is a 1/2 pixel point, then select an integer pixel point in the target pixel matrix as a reference pixel point;

If the type of the target interpolation position is a 1/4 pixel point or a 3/4 pixel point, a 1/2 pixel point in the target pixel matrix is selected as a reference pixel point.
The device according to claim 28, wherein the target filtering parameter is an N-dimensional parameter vector, and N is a positive integer; and,

The interpolation unit is further configured to:

Controlling N data selectors to select N pieces of data, the N pieces of data are respectively the data processing results after the target pixel matrix is processed by using the N parameter values in the target filtering parameters, wherein the N pieces of data are selected The device is pre-divided into the first group and the second group;

inputting the data outputted by the data selectors in the first group into the first adder, and inputting the data outputted by the data selectors in the second group into the second adder;

The data output by the first adder and the data output by the second adder are input to a subtractor, and the data output by the subtracter is used as the interpolation result of the target interpolation position.
An electronic device, comprising: a processor and a memory;

the memory for storing program instructions;

The processor executes the program instructions stored in the memory, and when the program instructions are executed, the processor is configured to perform the following steps:

Select the target interpolation position of the target image block;

From the preset mapping table, query the target pixel matrix offset corresponding to the target interpolation position and the target filtering parameter corresponding to the target interpolation position;

The target interpolation position is interpolated based on the target pixel matrix offset and the target filter parameter.
The device according to claim 41, wherein the mapping table includes first mapping information of the interpolation position and the offset of the pixel matrix;

The first mapping information is generated by the following steps:

obtaining an input pixel matrix corresponding to an image block of a specified size, wherein the size of the image block of the specified size is the same as that of the target image block;

determining the pixel matrix offset of the pixel matrix relative to the input pixel matrix required for each interpolation position in the image block of the specified size;

The first mapping information is generated based on the pixel matrix offset corresponding to each interpolation position.
The device according to claim 42, wherein the generating the first mapping information based on the pixel matrix offset corresponding to each interpolation position comprises:

setting a position number for each of the interpolation positions;

Based on the pixel matrix offset corresponding to each interpolation position, a mapping between the position number and the pixel matrix offset is established to obtain first mapping information.
The device according to claim 42, wherein the pixel matrix offset corresponding to each interpolation position comprises a horizontal pixel matrix offset and a vertical pixel matrix offset;

The first mapping information includes a mapping relationship between the interpolation positions and the offset of the pixel matrix in the horizontal direction, and includes a mapping relationship between the interpolation positions and the offset of the pixel matrix in the vertical direction.
The apparatus according to claim 41, wherein the mapping table includes second mapping information of interpolation positions and filtering parameters;

The second mapping information is generated through the following steps:

determining the type of each interpolation position in an image block of a specified size, wherein the size of the image block of the specified size is the same as the size of the target image block;

Based on the types of the interpolation positions, determine the filtering parameters corresponding to the interpolation positions;

The second mapping information is generated based on the filtering parameters corresponding to the interpolation positions.
The apparatus according to claim 45, wherein the generating the second mapping information based on the filtering parameters corresponding to the interpolation positions, comprises:

setting a position number for each of the interpolation positions;

Set the index number for each filter parameter;

Based on the filtering parameters corresponding to the interpolation positions, a mapping between position numbers and index numbers is established to obtain second mapping information.
The apparatus according to claim 45, wherein the types of the interpolation positions include the types of the interpolation positions in the horizontal direction and the types of the interpolation positions in the vertical direction.
The device according to claim 47, wherein the determining the filtering parameters corresponding to the interpolation positions based on the types of the interpolation positions comprises:

determining the filtering parameters corresponding to the interpolation positions in the horizontal direction based on the types of the interpolation positions in the horizontal direction;

Based on the types of the interpolation positions in the vertical direction, filter parameters corresponding to the interpolation positions in the vertical direction are determined.
The apparatus according to claim 47, wherein the second mapping information includes a mapping relationship between the interpolation positions and the filtering parameters corresponding to the horizontal direction, and includes the interpolation positions and the vertical The mapping relationship of the corresponding filter parameters in the direction.
The apparatus according to claim 45, wherein the type of interpolation position includes at least one of the following: 1/4 pixel point, 1/2 pixel point, 3/4 pixel point and whole pixel point.
The apparatus according to claim 45, wherein the filtering parameters corresponding to different types of interpolation positions are parameter vectors with the same dimension.
The device according to claim 50, wherein the filtering parameters corresponding to the 1/4 pixel points are determined by the following steps:

Obtain the first default filter parameter corresponding to the 1/4 pixel point, the first default filter parameter is an M-dimensional parameter vector, and M is a positive integer;

Add 0 to the end of the first default filtering parameter to obtain the final filtering parameter corresponding to the 1/4 pixel point, and the final filtering parameter is an N-dimensional parameter vector, N=M+1.
The apparatus according to claim 50, wherein the filtering parameter corresponding to the 1/2 pixel point is a second default filtering parameter, the second default filtering parameter is an M-dimensional parameter vector, and M is a positive integer.
The device according to claim 50, wherein the filtering parameters corresponding to the 3/4 pixel points are determined by the following steps:

Obtain the third default filter parameter corresponding to the 3/4 pixel point, the third default filter parameter is an M-dimensional parameter vector, and M is a positive integer;

0 is added before the first digit of the third default filter parameter to obtain the final filter parameter corresponding to the 3/4 pixel point, and the final filter parameter is an N-dimensional parameter vector, N=M+1.
The device according to claim 50, wherein the filtering parameter corresponding to the integer pixel is an M-dimensional parameter vector with a parameter value of 0, and M is a positive integer.
The device according to claim 50, wherein the filtering parameters corresponding to the 1/4 pixel points, the filtering parameters corresponding to the 1/2 pixel points, and the filtering parameters corresponding to the 3/4 pixel points The parameter value and the parameter value of the same dimension in the filtering parameter corresponding to the integer pixel are not less than 0 or are not greater than 0.
The apparatus of claim 41, wherein the processor is further configured to:

obtaining the input pixel matrix corresponding to the target image block;

Based on the target pixel matrix offset and the input pixel matrix, determine a target pixel matrix required for interpolation of the target interpolation position;

The target interpolation position is interpolated based on the target pixel matrix and the target filter parameter.
The apparatus of claim 47, wherein the processor is further configured to:

Based on the type of the target interpolation position, select the reference pixel point in the target pixel matrix;

Filter processing is performed on the reference pixel point based on the target filtering parameter to obtain an interpolation result of the target interpolation position.
The apparatus of claim 48, wherein the processor is further configured to:

If the type of the target interpolation position is a 1/2 pixel point, then select an integer pixel point in the target pixel matrix as a reference pixel point;

If the type of the target interpolation position is a 1/4 pixel point or a 3/4 pixel point, a 1/2 pixel point in the target pixel matrix is selected as a reference pixel point.
The device according to claim 48, wherein the target filtering parameter is an N-dimensional parameter vector, and N is a positive integer; and,

The processor is further used for:

Controlling N data selectors to select N pieces of data, the N pieces of data are respectively the data processing results after the target pixel matrix is processed by using the N parameter values in the target filtering parameters, wherein the N pieces of data are selected The device is pre-divided into the first group and the second group;

Inputting the data outputted by the data selectors in the first group to the first adder, and inputting the data outputted by the data selectors in the second group to the second adder;

The data output by the first adder and the data output by the second adder are input to a subtractor, and the data output by the subtracter is used as the interpolation result of the target interpolation position.
A computer-readable medium, characterized in that, a computer program is stored in the computer-readable medium, and when the computer program is executed by a processor, the method according to any one of claims 1-20 is implemented.