WO2023246700A1

WO2023246700A1 - Point cloud attribute encoding method, point cloud attribute decoding method, and storage medium

Info

Publication number: WO2023246700A1
Application number: PCT/CN2023/101081
Authority: WO
Inventors: 陈悦汝; 王静; 李革; 高文
Original assignee: 鹏城实验室
Priority date: 2022-06-20
Filing date: 2023-06-19
Publication date: 2023-12-28
Also published as: CN115278269A; CN115278269B

Abstract

Disclosed in the present invention are a point cloud attribute encoding method, a point cloud attribute decoding method, and a storage medium. The encoding method comprises: obtaining a target sorting code on the basis of point cloud data to be encoded, and then sorting said point cloud data; and sequentially performing attribute encoding according to the sorting, involving: calculating a difference distance between the current point to be encoded and the most adjacent point, comparing the difference distance with a threshold distance, adjusting the current quantization parameter or the current quantization rounding method according to a comparison result, and then quantizing an input coefficient by using the current adjusted quantization parameter or the current adjusted quantization rounding method, so as to obtain a quantized coefficient. During encoding, the difference distance between the current point to be encoded and the most adjacent point is calculated and then compared with the threshold distance, the current quantization parameter or the current quantization rounding method is adjusted according to the comparison result, and the input coefficient is then quantized to obtain the quantized coefficient, that is, a point with a relatively large prediction error is selected by means of the determination of the distance, and the quantization parameter is adjusted to reduce the prediction error, thereby improving the precision of reconstructed point cloud data.

Description

A point cloud attribute encoding method, point cloud attribute decoding method and storage medium

Technical field

The invention relates to the field of data processing technology, and in particular to a point cloud attribute encoding method, a point cloud attribute decoding method and a storage medium.

Background technique

The main steps of existing point cloud attribute encoding technology are:

1) Arrange the point cloud into a one-dimensional sequence in a certain order, and encode each point in sequence;

2) Use the information of previously encoded points to predict the attribute value of a certain point, and obtain the residual difference between the predicted value and the true attribute value;

3) Quantize the residual in 2) according to the quantization accuracy set in advance to obtain the residual quantization coefficient, and perform entropy coding on the residual quantization coefficient to complete the encoding of this point; if transformation technology is used, further quantize the residual in 2) The residual is transformed to obtain the transform coefficients, and then the transform coefficients are quantized to obtain the transform quantized coefficients. The transform quantized coefficients are entropy-encoded to obtain a code stream, and the encoding of this point is completed.

The main steps corresponding to the decoding technology are:

4) Perform corresponding entropy decoding on the code stream to obtain quantized coefficients, and perform inverse quantization on the quantized coefficients to obtain the reconstruction residual; if transformation technology is used, perform corresponding entropy decoding on the code stream to obtain quantized coefficients, and perform inverse quantization on the quantized coefficients. and inverse transformation to obtain the reconstruction residual.

5) Add the reconstruction residual to the predicted value in 2) to obtain the reconstructed attribute value, which is used for prediction of subsequent points.

However, existing prediction techniques make predictions based on the assumption that points with close geometric distances have high correlation. The prediction error is also large when the distance is large, and since subsequent points are predicted based on previously reconstructed points, there is also the defect of accumulation of quantization errors, resulting in a large overall average error in the decoded reconstructed point cloud and low accuracy.

Therefore, the existing technology still needs to be improved and improved.

Contents of the invention

The main purpose of the present invention is to provide a point cloud attribute encoding method, a point cloud attribute decoding method and a storage medium, aiming to solve the problem of poor accuracy in reconstructing point clouds in the prior art.

In order to achieve the above objects, the present invention adopts the following technical solutions:

A point cloud attribute encoding method, the point cloud attribute encoding method includes the following steps:

Obtain the target sorting code based on the point cloud data to be encoded according to the preset rules, and sort the point cloud data to be encoded according to the target sorting code to obtain the first sorted point cloud data; wherein the preset rule is the original Serial number arrangement rules, Morton order calculation rules or Hilbert order calculation rules, the point cloud data to be encoded are point cloud data whose attributes are to be encoded;

Starting from the Nth point in the first sorted point cloud data, calculate the difference distance between the current point to be encoded and the most adjacent point according to the preset distance calculation method; wherein the preset distance calculation method is Geometric distance calculation method, or a comprehensive distance calculation method composed of geometric distance and attribute distance;

Compare the difference distance and the threshold distance, adjust the current quantization parameter according to the comparison result, and obtain the current adjusted quantization parameter, or select different rounding intervals according to the comparison result, and obtain the current adjusted quantization rounding method;

After using the currently adjusted quantization parameter or the currently adjusted quantization rounding method to quantize the input coefficient to obtain the quantized coefficient, the quantized coefficient is entropy encoded to obtain a code stream; wherein the input coefficient includes a residual coefficient and Transformation coefficient.

In the point cloud attribute encoding method, the difference distance and the threshold distance are compared, and the current quantization parameter is adjusted according to the comparison result to obtain the current adjusted quantization parameter, or the current adjustment quantization parameter is selected according to the comparison result. Select different rounding intervals and obtain the current adjustment quantization rounding method. The steps include:

Compare the difference distance with the threshold distance;

If the difference distance is greater than the threshold distance, add the initial quantization parameter to the offset value to obtain the current adjusted quantization parameter, or select the first rounding interval as the current adjusted quantization rounding method;

If the difference distance is not greater than the threshold distance, the initial quantization parameter is selected as the currently adjusted quantization coefficient, or the second rounding interval is selected as the currently adjusted quantization rounding method.

In the point cloud attribute encoding method, after the input coefficient is quantized using the currently adjusted quantization parameter or the currently adjusted quantization rounding method to obtain the quantized coefficient, the quantized coefficient is entropy encoded to obtain the code stream. Steps include:

Predict the attribute value of the current point to be encoded to obtain a predicted value, calculate the residual coefficient between the predicted value and the real attribute value of the current point to be encoded; use the currently adjusted quantization parameter or the currently adjusted quantization rounding method to The residual quantized coefficients obtained by quantizing the residual coefficients are entropy encoded to obtain the code stream; wherein the input coefficients include the residual coefficients, and the quantized coefficients include the The residual quantization coefficient;

Or, predict the attribute value of the current point to be encoded to obtain the predicted value, calculate the residual coefficient between the predicted value and the real attribute value of the current point to be encoded, and transform the residual coefficient to obtain the transformation coefficient ; Using the transform quantization coefficient obtained by quantizing the transform coefficient using the currently adjusted quantization parameter or the currently adjusted quantization rounding method, performing entropy coding on the transform quantization coefficient to obtain the code stream; wherein, the The input coefficients include the transform coefficients, and the quantized coefficients include the transform quantized coefficients.

A computer-readable storage medium stores a point cloud attribute encoding program on the computer-readable storage medium. When the point cloud attribute encoding program is executed by a processor, the steps of the point cloud attribute encoding method as described above are implemented.

A point cloud attribute decoding method, the point cloud attribute decoding method includes the following steps:

Sort the point cloud data to be decoded according to preset rules to obtain the second sorted point cloud data; Wherein, the point cloud data to be decoded is point cloud data whose attributes are to be decoded;

Starting from the Nth point in the second sorted point cloud data, calculate the difference distance between the current point to be decoded and the most adjacent point according to the preset distance calculation method;

Compare the difference distance and the threshold distance, adjust the current inverse quantization parameter according to the comparison result, and obtain the current adjusted inverse quantization parameter;

Entropy decoding is performed on the encoded code stream to obtain reconstructed quantization coefficients, and the reconstructed quantization coefficients are inversely quantized according to the currently adjusted inverse quantization parameter to obtain reconstructed input coefficients.

In the point cloud attribute decoding method, the step of sorting the point cloud data to be decoded according to preset rules and obtaining the second sorted point cloud data specifically includes:

Obtain the target sorting code corresponding to the point cloud data to be decoded according to the original serial number arrangement rules, Morton order calculation rules or Hilbert order calculation rules; wherein the target sorting code is the input serial number or Morton code or Hill Bert code;

Sort the points in the point cloud data to be decoded in order from small to large or from large to small according to the target sorting code to obtain the second sorted point cloud data.

In the point cloud attribute decoding method, starting from the Nth point in the second sorted point cloud data, the difference distance between the current point to be decoded and the most adjacent point is calculated according to a preset distance calculation method. The steps specifically include:

According to the geometric distance calculation method, or the comprehensive distance calculation method composed of geometric distance and attribute distance, the point closest to the current point to be decoded is searched as the nearest neighbor point of the current point to be decoded; where N is a positive integer greater than 1;

Or, use the first m points of the current point to be decoded as the most adjacent points of the current point to be decoded;

The distance difference between the current point to be decoded and the most adjacent point is calculated as the difference distance.

In the point cloud attribute decoding method, the step of comparing the difference distance and the threshold distance, adjusting the current inverse quantization parameter according to the comparison result, and obtaining the current adjusted inverse quantization parameter includes:

Compare the difference distance with the threshold distance;

If the difference distance is greater than the threshold distance, add the offset value to the initial inverse quantization parameter to obtain the current adjusted inverse quantization parameter;

If the difference distance is not greater than the threshold distance, the initial inverse quantization parameter is selected as the current adjusted inverse quantization parameter.

In the point cloud attribute decoding method, the encoded code stream is entropy decoded to obtain reconstructed quantization coefficients, and the reconstructed quantization coefficients are inversely quantized according to the currently adjusted inverse quantization parameters to obtain the reconstructed input coefficients. include:

Perform entropy decoding on the encoded code stream to obtain transform reconstruction quantization coefficients or residual reconstruction quantization coefficients;

Using the currently adjusted inverse quantization parameter to perform inverse quantization on the transform reconstruction quantization coefficient or the residual reconstruction quantization coefficient, correspondingly obtain the transform reconstruction coefficient or the residual reconstruction coefficient; wherein the reconstruction input coefficient includes the transform reconstruction coefficient reconstruction coefficients and the residual reconstruction coefficients.

A computer-readable storage medium stores a point cloud attribute decoding program on the computer-readable storage medium. When the point cloud attribute decoding program is executed by a processor, the steps of the point cloud attribute decoding method as described above are implemented.

Compared with the existing technology, the present invention provides a point cloud attribute encoding method, a point cloud attribute decoding method and a storage medium. The encoding method includes: obtaining a target sorting code based on the point cloud data to be encoded according to preset rules, and according to the target The sorting code sorts the point cloud data to be encoded; performs attribute encoding on the points in the point cloud data to be encoded in order: calculates the difference distance between the current point to be encoded and the most adjacent point, and compares it with the threshold distance. According to the comparison As a result, adjust the current quantization parameter to get the current adjusted quantization parameter or select the rounding interval to get the current adjusted quantization rounding method; quantize the input coefficient according to the current adjusted quantization parameter or the current adjusted quantization rounding method to obtain the quantization coefficient; perform entropy on the quantization coefficient Encode to get the code stream. When encoding point cloud data, calculate the difference distance between the current point to be encoded and the most adjacent point, compare it with the threshold distance, and adjust the current quantization parameter or select the rounding interval according to the comparison result to obtain the current adjusted quantization rounding. rounding method, and then use the current adjusted quantization parameter or the current adjusted quantization rounding method to perform the input coefficient The quantization coefficient is obtained by quantification, that is, by judging the distance, screening points with large prediction errors, adjusting quantization parameters or quantification rounding methods, reducing the prediction error, thereby effectively improving the accuracy of reconstructed point cloud data.

Description of the drawings

Figure 1 is a flow chart of a preferred embodiment of the point cloud attribute encoding method provided by the present invention;

Figure 2 is a schematic diagram of the method for determining the nearest adjacent point in a preferred embodiment of the point cloud attribute encoding method provided by the present invention;

Figure 3 is a flow chart of step S30 in a preferred embodiment of the point cloud attribute encoding method provided by the present invention;

Figure 4 is a flow chart of step S40 in a preferred embodiment of the point cloud attribute encoding method provided by the present invention;

Figure 5 is another flow chart of step S40 in the preferred embodiment of the point cloud attribute encoding method provided by the present invention;

Figure 6 is a flow chart of a preferred embodiment of the point cloud attribute decoding method provided by the present invention;

Figure 7 is a flow chart of step S100 in a preferred embodiment of the point cloud attribute encoding method provided by the present invention;

Figure 8 is a flow chart of step S200 in a preferred embodiment of the point cloud attribute encoding method provided by the present invention;

Figure 9 is a flow chart of step S300 in a preferred embodiment of the point cloud attribute encoding method provided by the present invention;

Figure 10 is a flow chart of step S400 in a preferred embodiment of the point cloud attribute encoding method provided by the present invention.

Detailed ways

In order to make the purpose, technical solution and effect of the present invention clearer and clearer, refer to the accompanying drawings below. The present invention will be further described in detail with reference to examples. It should be understood that the specific embodiments described here are only used to explain the present invention and are not intended to limit the present invention.

Those skilled in the art will understand that, unless expressly stated otherwise, the singular forms "a", "an", "the" and "the" used herein may also include the plural form. It should be further understood that the word "comprising" used in the description of the present invention refers to the presence of stated features, integers, steps, operations, elements and/or components, but does not exclude the presence or addition of one or more other features, Integers, steps, operations, elements, components and/or groups thereof. It will be understood that when we refer to an element being "connected" or "coupled" to another element, it can be directly connected or coupled to the other element or intervening elements may also be present. Additionally, "connected" or "coupled" as used herein may include wireless connections or wireless couplings. As used herein, the term "and/or" includes all or any unit and all combinations of one or more of the associated listed items.

It will be understood by those skilled in the art that, unless otherwise defined, all terms (including technical terms and scientific terms) used herein have the same meaning as commonly understood by one of ordinary skill in the art to which this invention belongs. It should also be understood that terms, such as those defined in general dictionaries, are to be understood to have meanings consistent with their meaning in the context of the prior art, and are not to be used in an idealistic or overly descriptive manner unless specifically defined as here. to explain the formal meaning.

The technical solutions in the embodiments of the present invention are clearly and completely described below with reference to the accompanying drawings of the embodiments of the present invention. Obviously, the described embodiments are only some of the embodiments of the present invention, rather than all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those of ordinary skill in the art without creative efforts fall within the scope of protection of the present invention.

Many specific details are set forth in the following description to fully understand the present invention. However, the present invention can also be implemented in other ways different from those described here. Those skilled in the art can do so without departing from the connotation of the present invention. Similar generalizations are made, and therefore the present invention is not limited to the specific embodiments disclosed below.

With the rapid development of science and technology, especially the development of computer technology, three-dimensional reconstruction and other technologies The technology has been widely used in various fields such as architectural design, game development, and cultural relic protection. Among them, point cloud compression encoding and decoding technology is one of the key technologies for 3D reconstruction. Point clouds are obtained by sampling object surfaces with 3D scanning equipment. Each point cloud may include various attribute information, such as color information, reflectivity information, etc. The purpose of point cloud compression encoding and decoding is to remove redundancy as much as possible while retaining the original attribute information of massive point cloud data and improve system storage and transmission efficiency.

The invention provides a point cloud attribute encoding method, a point cloud attribute decoding method and a storage medium. In this application, when encoding point cloud data, after calculating the difference distance between the current point to be encoded and the most adjacent point, it is compared with the threshold distance, and the current inverse quantization parameter is adjusted according to the comparison result, or different rounding is selected. The interval is used as the current quantization rounding method, and then the obtained currently adjusted inverse quantization parameter or the currently adjusted quantization rounding method is used to quantize and entropy encode the input coefficients to obtain the code stream; and by decoding the point cloud data, Similarly, after calculating the difference distance between the current point to be encoded and the most adjacent point, compare it with the threshold distance, adjust the current inverse quantization parameter according to the comparison result, and then use the obtained current adjusted inverse quantization parameter to encode the code stream after entropy decoding. Inverse quantization is performed to obtain the reconstruction input coefficient; that is, during encoding and decoding, error points are screened out by judging the distance, and the quantization parameters are adjusted, thereby reducing the defect of accumulation of quantization errors during prediction, which results in decoding and reconstructing point clouds. The problem of increased overall average error effectively improves the accuracy of reconstructed point cloud data.

The design solution of the point cloud attribute encoding method is described below through specific exemplary embodiments. It should be noted that the following embodiments are only used to explain the technical solution of the invention and are not specifically limited:

Please refer to Figure 1. A point cloud attribute encoding method provided by the present invention includes the following steps:

S10. Obtain the target sorting code based on the point cloud data to be encoded according to the preset rules, and sort the point cloud data to be encoded according to the target sorting code to obtain the first sorted point cloud data. The first sorted point cloud The data is a one-dimensional sequence. Wherein, the preset rules are original sequence number arrangement rules, Morton order calculation rules or Hilbert order calculation rules, and the point cloud data to be encoded For point cloud data whose attributes are to be encoded, the original sequence number arrangement rule refers to a rule that adopts the original input sequence number of the point cloud data to be encoded when it is input.

Among them, point cloud data refers to a set of vectors in a three-dimensional coordinate system. In this embodiment, the point cloud data to be encoded may be point cloud data obtained by scanning, such as a laser radar scanning point cloud, or a point cloud using VR, etc. Each point cloud is recorded in the form of a point, and each point contains There are three-dimensional coordinates and attribute information (such as color information and reflectance information).

Specifically, when encoding the point cloud data to be encoded, first, the point cloud data to be encoded is converted into multiple target sorting codes according to the original sequence number arrangement rules, Morton order calculation rules or Hilbert order calculation rules. , and then sort the multiple target sorting codes in order from small to large or from large to small to obtain the first sorted point cloud data, thereby achieving effective sorting of the point cloud data to be encoded, so as to facilitate the first sorted point cloud data. A sorted point cloud data is encoded sequentially.

S20. Starting from the Nth point in the first sorted point cloud data, calculate the difference distance between the current point to be encoded and the most adjacent point according to the preset distance calculation method; where N is a positive value greater than 1. Integer, the preset distance calculation method is a geometric distance calculation method, or a comprehensive distance calculation method composed of geometric distance and attribute distance.

Specifically, the geometric distance may be the Euclidean distance, that is, the distance on the x, y, and z axes between two points (such as the current point A to be encoded and the encoded point P) is calculated, denoted as Dg. In some cases point clouds may have multiple attributes, such as color attributes and reflectance attributes. After the reflectance attribute has been completed, the reconstructed reflectance attribute values of all points can be obtained. Then when encoding color, the attribute distance can be equal to the absolute value of the difference between the reflectivity attribute reconstruction values of point A and point P, recorded as Da, then the comprehensive distance Dc=Dg+Da*C, where C is Adjustable parameters.

Then, after the point cloud data to be encoded is sorted, starting from the Nth point in the first sorted point cloud data, the point cloud data to be encoded is encoded in sequence according to the first sorted point cloud data. The encoding process is as follows :

The difference distance D between the current point to be encoded and the most adjacent point is calculated using a geometric distance calculation method or a comprehensive distance calculation method composed of geometric distance and attribute distance. It can also be based on the In the coding sorting of the first sorted point cloud data, the previous point of the current point to be coded is directly regarded as the most adjacent point, and the difference distance between the current point to be coded and the most adjacent point is calculated, so that it can be based on a variety of This method accurately calculates the difference distance between the current point to be encoded and the most adjacent point, which serves as a basis for adjusting the current quantization parameters or the current quantization rounding method. Or the sorting distance is directly used as the difference distance to adjust the current quantization parameter or the current quantization rounding method. The sorting distance of two point cloud data points is the sum of the number of points between the two points of the first sorted point cloud data. 1. Sorting distance is the distance calculation method with the lowest computational complexity.

The method for determining the nearest adjacent point can be found in Figure 2:

1. Determine the nearest adjacent point based on geometric distance:

Based on the known coordinate information, calculate the Euclidean distance (or Manhattan distance, the square of the Euclidean distance, etc.) between data points 1-7 (coded points) and point 8 (current point to be coded), and compare to obtain the distance The smallest data point is used as the nearest neighbor point to point 8.

2. Determine the nearest neighbor point based on the comprehensive distance composed of geometric distance and attribute distance:

Based on the known coordinate information and other known attribute information, calculate the comprehensive distance between data points 1-7 and point 8, and compare the data point with the smallest distance as the nearest neighbor point of point 8. For example, if the currently encoded attribute is the color attribute and the known attribute is the reflectance attribute, then the comprehensive distance Dc is: Dc=Dg+Da*C. Among them, C is an adjustable parameter; Dg is the Euclidean distance; Da is the distance of the reflectivity attribute, which is equal to the absolute value of the difference between the reconstructed values of the reflectance attribute of points 1-7 and point 8.

3. Determine the nearest neighbor point based on the sorting distance:

The previous point of point 8 (data point No. 7) is directly regarded as the nearest neighbor point, and the nearest sorting distance is: 8-7=1.

S30. Compare the difference distance and the threshold distance, adjust the current quantization parameter according to the comparison result to obtain the current adjusted quantization parameter, or select different rounding intervals according to the comparison result to obtain the current adjusted quantization rounding method.

Specifically, after calculating the difference distance between the current point to be encoded and the most adjacent point, compare The difference distance and the threshold distance T, and adjust the current quantization parameter according to the comparison result to obtain the current adjustment quantization parameter, or select different rounding intervals according to the comparison result to obtain the current adjustment quantization rounding method, realizing the current adjustment quantization rounding method according to the comparison result. The above preset thresholds are used to adjust the previous quantization parameters or the current quantization rounding method accordingly, thereby filtering out error points and improving the accuracy of reconstructing point clouds; and by customizing the right shift rounding interval, more smaller quantization coefficients can be obtained , thereby reducing the coding length. Although it will introduce a larger average quantization error, it balances the advantages and disadvantages, thereby improving the overall coding efficiency.

Further, please refer to Figure 3. S30: Compare the difference distance and the threshold distance, adjust the current quantization parameter according to the comparison result, and obtain the current adjusted quantization parameter, or select different rounding intervals according to the comparison result to obtain the current quantization parameter. The steps to adjust the quantization rounding method include:

S31. Compare the difference distance with the threshold distance;

S32. If the difference distance is greater than the threshold distance, add the offset value to the initial quantization parameter to obtain the current adjusted quantization parameter, or select the first rounding interval as the current adjusted quantization rounding method;

S33. If the difference distance is not greater than the threshold distance, select the initial quantization parameter as the currently adjusted quantization coefficient, or select the second rounding interval as the currently adjusted quantization rounding method. Wherein, the first rounding interval may be rounding to 60s, and the second rounding interval may be rounding to 70s.

Specifically, after calculating the difference distance between the current point to be encoded and the most adjacent point, the difference distance is compared with the threshold distance T, and the user can choose to adjust the current quantization parameter according to the comparison result: if When the difference distance D is greater than the threshold distance T, that is, when D>T, add a user-preset offset value QP _shift to the initial quantization parameter QP _old to obtain the current adjusted quantization parameter QP _new : QP _new = QP _old + QP _shift , where QP _new is a positive integer greater than 0; and if the difference distance D is not greater than the threshold distance T, that is, when D <= T, the initial quantization parameter is selected as the current adjusted quantization coefficient;

Alternatively, you can also choose to change the quantification method based on the comparison results: because the usual quantification method is uniform quantization, that is, the current quantization coefficient = round (input coefficient/current quantization parameter), where round is the rounding interval, because in this case, the input coefficient and the current quantization parameter are unchanged, so it can be passed Change the rounding interval to change the quantization method, specifically: if the difference distance is greater than the threshold distance, improve the accuracy of the rounding interval, and select the first rounding interval (rounding) as the current Quantization rounding method; if the difference distance is not greater than the threshold distance, select the second rounding interval (rounding) as the current quantization rounding method, so that when the difference distance is greater than the threshold distance When the threshold distance is used, selecting a rounding interval with greater rounding accuracy will help improve the accuracy of constructing and reconstructing unit data.

Among them, T is related to the specific distribution of the point cloud and can be calculated according to the following formula:

Among them, maxSize is the average (or maximum) side length of the minimum bounding box after the point cloud geometric coordinates are fixed-pointed, that is, a cube is used to wrap the point cloud. The cube size is the smallest, N is the number of point cloud geometric points, a is an adjustable parameter. For example, a is set to 12, that is, T is the diagonal Manhattan distance of the macroblock where point A is located, and the side length of the macroblock is It should be noted that the difference distance and the threshold distance are compared, and the current quantization parameter is adjusted according to the comparison result, or the difference distance and the threshold distance are compared, and different rounding intervals are selected according to the comparison result. , the two threshold distances can be the same value or different values.

Another method for determining the distance threshold T in practical applications. It is assumed that the macroblock side length dimension can be expressed as an exponent of 2. T is calculated according to the following formula:
L ₀ =3*(Log2(maxSize)-((Log2(N/4))/2));
T＝(2^(L ₀ /3))*b;

Among them, b is an adjustable parameter, and b can be equal to 6 in the application. Further, alternatively, L ₀ can be dynamically adjusted, thereby adjusting T. Determine the initial L (initial state: L=L ₀ ) according to the above formula, and then to dynamically adjust the size of L. The specific rule is to count the average number BN of the number of points in KN point cloud groupings. If the value of BN is less than BN ₁ , then L=L+1; if the value of BN is greater than BN ₂ , then L=L- 1; otherwise, L remains unchanged (BN ₁ and BN ₂ are variable parameters, BN ₁ <BN ₂ , BN ₁ and BN ₂ are positive integers greater than 0). For each KM point cloud grouping, a judgment and adjustment are made (KN and KM are variable parameters, KN<KM, KN and KM are positive integers greater than 1. KM is generally a multiple of KN). Finally, according to the adjusted L, T=(2^(L/3))*b is calculated.

After calculating the difference distance between the current point to be encoded and the most adjacent point, the difference distance is compared with the threshold distance, that is, the distance is determined, and the current quantization parameter or the current quantization parameter is adjusted according to the comparison result. Quantitative rounding method, thereby effectively screening out error points and improving the accuracy of reconstructing point clouds.

Further, please continue to refer to Figure 1. S40: After quantizing the input coefficient using the currently adjusted quantization parameter or the currently adjusted quantization rounding method to obtain a quantized coefficient, entropy coding is performed on the quantized coefficient to obtain a code stream; Wherein, the input coefficients include residual coefficients and transformation coefficients.

Specifically, after obtaining the currently adjusted quantization parameter or the currently adjusted quantization rounding method, the residual coefficient or the transform coefficient is quantized using the currently adjusted quantization parameter or the currently adjusted quantization rounding method to obtain a quantized coefficient. Finally, the entropy coding operation is performed on the quantization coefficient to obtain the code stream, and the reconstructed point cloud data is obtained at the same time. The encoding of the current point to be encoded is completed, so that the residual can be calculated according to the adjusted current quantization parameter or the current quantization rounding method. After quantization and entropy coding of the difference coefficients or transformation coefficients, more accurate reconstructed point cloud data and code streams can be obtained.

Further, please refer to Figure 4. In S40, after quantizing the input coefficient using the currently adjusted quantization parameter or the currently adjusted quantization rounding method to obtain a quantized coefficient, entropy encoding is performed on the quantized coefficient to obtain a code. The flow steps include:

S41. Predict the attribute value of the current point to be encoded to obtain a predicted value, and calculate the residual coefficient between the predicted value and the true attribute value of the current point to be encoded;

S42. Use the currently adjusted quantization parameter or the currently adjusted quantization rounding method to calculate the Residual quantization coefficients obtained by further quantization of residual coefficients are entropy-encoded to obtain the code stream; wherein the input coefficients include the residual coefficients, and the quantization coefficients include the residual Differential quantization coefficient;

Or please refer to Figure 5 , the step of S40, using the currently adjusted quantization parameter or the currently adjusted quantization rounding method to quantize the input coefficient to obtain the quantized coefficient, and then perform entropy coding on the quantized coefficient to obtain a code stream. include:

A41. Predict the attribute value of the current point to be encoded to obtain the predicted value, calculate the residual coefficient between the predicted value and the real attribute value of the current point to be encoded, and transform the residual coefficient to obtain the transformation coefficient ;

A42. Use the transform quantization coefficient obtained by quantizing the transform coefficient using the currently adjusted quantization parameter or the currently adjusted quantization rounding method, and perform entropy coding on the transform quantization coefficient to obtain the code stream; wherein, The input coefficients include the transform coefficients, and the quantized coefficients include the transform quantized coefficients.

Specifically, when encoding the current point to be encoded, the attribute value of the current point to be encoded is predicted using the information of the previous encoded point to obtain the predicted value (fitting value).

For example, a general prediction value calculation method: assuming that the current point to be encoded is i, in the set of points that have been encoded, find the three neighbor points closest to the geometric distance (such as Euclidean distance) from i, then, the attributes of the current point to be encoded i The predicted value is the weighted average of the reconstructed attribute values of three neighbor points. Among them, the weight can be the reciprocal of the distance from the neighbor point to i. Therefore, the geometric information of the current point to be encoded, as well as the geometric information and attribute information of neighbor points are generally used to predict the attribute value of the current point to be encoded.

Next, the user can choose to use the residual coefficient or the transformation coefficient: when choosing to use the residual coefficient, the predicted value and the real attribute value of the current point to be encoded (input related to point cloud data) are calculated. the value recorded in the text); then, use the obtained currently adjusted quantization parameter or the currently adjusted quantization rounding method to further quantize the residual coefficient to obtain the residual quantization coefficient, and finally The residual quantization coefficient is entropy coded to obtain the code stream, and obtain reconstructed point cloud data at the same time, thereby completing the encoding of the current point to be encoded.

When the transform coefficient is selected to be used, the residual coefficient between the predicted value of the current point to be encoded and the real attribute value of the current point to be encoded is calculated, and the residual coefficient between the predicted value of the subsequent point and the subsequent real attribute value is further calculated. Residual coefficient, perform binary transformation on the two obtained residual coefficients to obtain the transformation coefficient. For example: assuming that the current point to be encoded is i, calculate the residual value (residual coefficient) R _i of i; continue processing point i+1, and calculate the residual value R _{i+1 of i+1} . The residual vector (R _i , R _i+1 ) can be obtained, the residual vector can be subjected to a binary transformation, and the transformation output can obtain the transformation coefficient (C _i , C _i+1 ). Then points i+2 and i+3 can be processed in the same way, and so on. K-element transformation can also be performed, which is not limited to binary transformation.

When quantizing the transform coefficients of the K-ary transform, the currently adjusted quantization parameter corresponding to a certain point of transformation or the currently adjusted quantization rounding method can be used to quantize all the transform coefficients of the K-ary transform. For example, the current point i to be encoded and the subsequent point i+1 are used as a group to perform binary transformation, and the current adjusted quantization parameter obtained from point i+1 or the currently adjusted quantization rounding method is used to transform the transformation coefficients C _i and C _i+1 for quantification.

Furthermore, in the same way, the transform coefficient is quantized using the currently adjusted quantization parameter or the currently adjusted quantization rounding method to obtain the transform quantized coefficient, and entropy coding is performed on the transform quantized coefficient to obtain the same result. The code stream is described and the reconstructed point cloud data is obtained at the same time, thereby completing the encoding of the current point to be encoded.

Wherein, the point cloud attribute data is quantified using the currently adjusted quantization parameter or the currently adjusted quantization rounding method. The point cloud attribute data may be three channels of point cloud color attributes (for example, R, G, B three channels). channel), or it can be one or two channels of the color attribute (such as the two channels of chroma channels Cb and Cr).

Further, in the embodiment of the present invention, based on the AVS-PCC PCRM software v7.0 version [1], the experimental results of comparing the point cloud attribute encoding method in this application with the original method of the platform were tested. The results are as shown in Table 1- As shown in Table 2.

Table 1

Table 2

Among them, Table 1 is a comparison table of rate distortion data of brightness, chromaticity and reflectivity under limited lossy geometry and lossy attributes; Table 2 is a comparison table of brightness, chroma and reflectance under lossless geometry and lossy attributes. Rate-distortion data comparison table; according to the data in Table 1-2, compared with the benchmark results of the test platform PCRM, under the conditions of limited lossy geometry and lossy attributes, under the conditions of lossless geometry and lossy attributes, for For the brightness attribute, the end-to-end attribute rate distortion of the present invention slightly increases by 0.5% and 0.4% respectively; for the chroma Cb attribute, the end-to-end attribute rate distortion of the present invention significantly reduces by 9.3% and 12.9% respectively; for chroma Cr attribute, the end-to-end attribute rate distortion of the present invention is significantly reduced by 13.2% and 16.7% respectively; the point cloud attribute encoding method provided by the embodiment of the present invention effectively reduces the end-to-end attribute rate distortion and has better encoding effect. .

Further, the present invention provides a computer-readable storage medium. The computer-readable storage medium stores a point cloud attribute encoding program. When the point cloud attribute encoding program is executed by a processor, the point cloud attributes as described above are implemented. The steps of the encoding method; since the steps of the point cloud attribute encoding method are described in detail above, they will not be described again here.

Further, please refer to Figure 6. The present invention provides a point cloud attribute decoding method. The point cloud attribute decoding method includes the following steps:

S100. Sort the point cloud data to be decoded according to preset rules to obtain the second sorted point cloud data. Wherein, the preset rules are original sequence number arrangement rules, Morton order calculation rules or Greek Albert order calculation rules, the point cloud data to be decoded are point cloud data whose attributes are to be decoded, and the original sequence number arrangement rule refers to the rule that uses the original input sequence number of the point cloud data to be encoded when it is input.

Specifically, when decoding the point cloud data to be decoded, in the same way, first, the point cloud data to be decoded is encoded according to the same preset rules (original sequence number arrangement rules, Morton order calculation rules or Hilber's Special order calculation rules) are converted into multiple target sorting codes, and then the multiple target sorting codes are sorted according to the same sorting method (from small to large or from large to small) during encoding to obtain the second sorted point cloud data. , thereby achieving effective sorting of the point cloud data to be decoded, so as to facilitate sequential decoding according to the second sorted point cloud data.

It should be noted that since the encoded point cloud data is decoded and the same preset rules and sorting methods are used to sort the point cloud data to be decoded, the first sorted point cloud data and the The second sorting of point cloud data is the same, here only to distinguish encoding sorting and decoding sorting.

Further, please refer to Figure 7. In S100, sorting point cloud data to be decoded according to preset rules, the step of obtaining the second sorted point cloud data specifically includes:

S110. Obtain the target sorting code corresponding to the point cloud data to be decoded according to the original serial number arrangement rules, Morton order calculation rules or Hilbert order calculation rules; wherein the target sorting code is the input serial number or Morton code or Hilbert code;

S120: Sort the points in the point cloud data to be decoded in order from small to large or from large to small according to the target sorting code to obtain the second sorted point cloud data.

Specifically, similarly, when decoding the point cloud data to be decoded, first, the point cloud data to be decoded is encoded according to the same preset rules (original sequence number arrangement rules, Morton order calculation rules or Hilber's Special order calculation rules) are converted into multiple target sorting codes, and then the multiple target sorting codes are sorted according to the same preset order (from small to large or from large to small) during encoding to obtain the second sorting point cloud. data. Specifically, the three-dimensional coordinates of the current points to be decoded are input in sequence, and a number (called the target sorting code) is output accordingly according to the Morton order calculation rules or the Hilbert order calculation rules. (including Morton code or Hilbert code, etc.), and then rearrange the points in order from small to large or from large to small according to the size of the output number, to obtain the second sorted point cloud data.

Further, please continue to refer to Figure 6. S200: Starting from the Nth point in the second sorted point cloud data, calculate the difference distance between the current point to be decoded and the most adjacent point according to the preset distance calculation method. .

Specifically, in the same way, after the point cloud data to be decoded is sorted, starting from the Nth point in the second sorted point cloud data, the point cloud data to be decoded is decoded in sequence according to the second sorted point cloud data. , the decoding process is as follows:

The difference distance between the current point to be decoded and the most adjacent point is calculated using the geometric distance calculation method, or a comprehensive distance calculation method composed of geometric distance and attribute distance. It can also be based on the decoding order of the second sorted point cloud data, directly treating the first m points of the current to-be-decoded point as the most adjacent points, and calculating the difference between the current to-be-decoded point and the most adjacent point. distance, so that the difference distance between the current point to be decoded and the most adjacent point can be accurately calculated according to various methods, which serves as a basis for adjusting the current quantization parameters or the current quantization rounding method.

Further, please refer to Figure 8. S200: Starting from the Nth point in the second sorted point cloud data, calculate the difference between the current point to be decoded and the most adjacent point according to the preset distance calculation method. The steps to value distance specifically include:

S210. According to the geometric distance calculation method, or the comprehensive distance calculation method composed of geometric distance and attribute distance, search for the point closest to the current point to be decoded as the closest point to the current point to be decoded;

S220, or, use the first m points of the current point to be decoded as the most adjacent points of the current point to be decoded;

S230. Calculate the distance difference between the current point to be decoded and the most adjacent point as the difference distance.

Specifically, after the point cloud data to be decoded is sorted, starting from the Nth point in the second sorted point cloud data, the point cloud data to be decoded is sequentially decoded according to the second sorted point cloud data. The decoding process as follows:

The difference distance D between the current point to be decoded and the most adjacent point is calculated using a geometric distance calculation method or a comprehensive distance calculation method composed of geometric distance and attribute distance. It is also possible to directly regard the first m points of the current point to be decoded as the most adjacent points according to the decoding order of the second sorted point cloud data. The specific method of determining the most adjacent points is similar to the determination in the encoding process. The most adjacent point method finally calculates the difference distance between the current point to be decoded and the most adjacent point.

Further, please continue to refer to Figure 6. S300: Compare the difference distance and the threshold distance, adjust the current inverse quantization parameter according to the comparison result, and obtain the current adjusted inverse quantization parameter.

Specifically, after calculating the difference distance between the current point to be decoded and the most adjacent point, the difference distance is compared with the threshold distance, and the current inverse quantization parameter is adjusted according to the comparison result to obtain the current adjusted inverse quantization parameters, effectively adjusting the pre-inverse quantization parameters according to the preset threshold, thereby filtering out error points and improving the accuracy of reconstructing point clouds; because the adjustment of the rounding interval does not affect the decoding process, there is no need to rely on the comparison results Select to change the current quantization rounding method.

Further, please refer to Figure 9. The step of S300, comparing the difference distance and the threshold distance, adjusting the current inverse quantization parameter according to the comparison result, and obtaining the current adjusted inverse quantization parameter includes:

S310. Compare the difference distance with the threshold distance;

S320. If the difference distance is greater than the threshold distance, add the offset value to the initial inverse quantization parameter to obtain the current adjusted inverse quantization parameter; wherein the initial inverse quantization parameter is equal to the initial quantization parameter;

S330. If the difference distance is not greater than the threshold distance, select the initial inverse quantization parameter as the current adjusted inverse quantization parameter.

Specifically, after calculating the difference distance between the current point to be decoded and the most adjacent point, the difference distance is compared with the threshold distance, and the current inverse quantization parameter is selected to be adjusted according to the comparison result: if When the difference distance D is greater than the threshold distance T, that is, when D>T, add the same offset value QP _shift to the initial inverse quantization parameter QP' _old to obtain the current adjusted inverse quantization parameter QP' _new : QP' _new =QP' _old +QP _shift ; and if the difference distance D is not greater than the threshold distance T, that is, when D<=T, the initial inverse quantization parameter is selected as the current adjusted inverse quantization parameter; where, Since the initial inverse quantization parameter is equal to the initial quantization parameter and the offset value is the same, the current adjusted inverse quantization parameter is equal to the current adjusted quantization parameter.

Further, please continue to refer to Figure 6. S400: Perform entropy decoding on the encoded code stream to obtain reconstructed quantization coefficients, and perform inverse quantization on the reconstructed quantization coefficients according to the currently adjusted inverse quantization parameters to obtain reconstructed input coefficients.

Specifically, after the current point cloud data to be encoded is encoded, the code stream is obtained. When decoding, the encoded code stream is entropy decoded to obtain reconstructed quantization coefficients, and the reconstructed quantization coefficients are obtained according to the currently adjusted inverse quantization parameters. The reconstructed quantized coefficients are inversely quantized to obtain the reconstructed input coefficients, thereby achieving reconstruction and restoration of the encoded code stream to obtain the reconstructed input coefficients, that is, transform reconstruction coefficients or residual reconstruction coefficients.

Further, please refer to Figure 10. In S400, perform entropy decoding on the encoded code stream to obtain reconstructed quantization coefficients, and perform inverse quantization on the reconstructed quantization coefficients according to the currently adjusted inverse quantization parameters to obtain the reconstructed input coefficients. The specific steps include:

S410. Perform entropy decoding on the encoded code stream to obtain the transform reconstruction quantization coefficient or the residual reconstruction quantization coefficient;

S420. Use the currently adjusted inverse quantization parameter to perform inverse quantization on the transform reconstruction quantization coefficient or the residual reconstruction quantization coefficient, and obtain the transform reconstruction coefficient or the residual reconstruction coefficient accordingly; wherein the reconstruction input coefficient includes the the transform reconstruction coefficients and the residual reconstruction coefficients.

Specifically, after the current point cloud data to be encoded is encoded, the code stream is obtained. When decoding, the encoded code stream is entropy decoded to obtain the transform reconstruction quantization coefficient or the residual reconstruction quantization coefficient; then, Using the currently adjusted inverse quantization parameter to perform inverse quantization on the residual reconstruction quantization coefficient to obtain a transform reconstruction coefficient, and performing inverse discrete cosine transform and prediction on the transform reconstruction coefficient to obtain a reconstruction attribute value; or, using the The inverse quantization parameter is currently adjusted to inversely quantize the residual reconstruction quantization coefficient to obtain the residual reconstruction coefficient, and the residual reconstruction coefficient is predicted to obtain the reconstruction attribute value, thereby realizing the encoding of the After entropy decoding and inverse quantization operations are performed on the current point cloud data to be encoded, the reconstructed input system is restored number, and finally restore the reconstructed attribute value, that is, reconstructed point cloud data.

Further, the present invention provides a computer-readable storage medium. The computer-readable storage medium stores a point cloud attribute decoding program. When the point cloud attribute decoding program is executed by a processor, the point cloud attributes as described above are implemented. The steps of the decoding method; since the steps of the point cloud attribute decoding method have been described in detail above, they will not be described again here.

To sum up, the present invention provides a point cloud attribute encoding method, a point cloud attribute decoding method and a storage medium. The encoding method includes: obtaining a target sorting code based on the point cloud data to be encoded according to preset rules, and according to the target The sorting code sorts the point cloud data to be encoded; performs attribute encoding on the points in the point cloud data to be encoded in order: calculates the difference distance between the current point to be encoded and the most adjacent point, and compares it with the threshold distance. According to the comparison As a result, adjust the current quantization parameter to get the current adjusted quantization parameter or select the quantization rounding interval to get the current adjusted quantization rounding method; quantize the input coefficient according to the current adjusted quantization parameter or the current amount adjustment rounding method to obtain the quantization coefficient; Perform entropy coding to obtain the code stream. When encoding point cloud data, calculate the difference distance between the current point to be encoded and the most adjacent point, compare it with the threshold distance, and adjust the current quantization parameter or select the rounding interval according to the comparison result to obtain the current adjusted quantization rounding. input method, and then use the obtained currently adjusted quantization parameters or the currently adjusted quantization rounding method to quantize the input coefficients to obtain the quantization coefficients, that is, by judging the distance, screen the points with large prediction errors, and adjust the quantization parameters or the quantization rounding method. , reducing the prediction error, thereby effectively improving the accuracy of reconstructed point cloud data.

It is understood that those of ordinary skill in the art can make equivalent substitutions or changes based on the technical solutions and inventive concepts of the present invention, and all such changes or substitutions should fall within the protection scope of the appended claims of the present invention.

Claims

A point cloud attribute encoding method, characterized in that the point cloud attribute encoding method includes the following steps:

Obtain the target sorting code based on the point cloud data to be encoded according to the preset rules, and sort the point cloud data to be encoded according to the target sorting code to obtain the first sorted point cloud data; wherein the preset rule is the original Serial number arrangement rules, Morton order calculation rules or Hilbert order calculation rules, the point cloud data to be encoded are point cloud data whose attributes are to be encoded;

Starting from the Nth point in the first sorted point cloud data, calculate the difference distance between the current point to be encoded and the most adjacent point according to the preset distance calculation method; where N is a positive integer greater than 1, The preset distance calculation method is a geometric distance calculation method, or a comprehensive distance calculation method composed of geometric distance and attribute distance;

Compare the difference distance and the threshold distance, adjust the current quantization parameter according to the comparison result, and obtain the current adjusted quantization parameter, or select different rounding intervals according to the comparison result, and obtain the current adjusted quantization rounding method;

After using the currently adjusted quantization parameter or the currently adjusted quantization rounding method to quantize the input coefficient to obtain the quantized coefficient, the quantized coefficient is entropy encoded to obtain a code stream; wherein the input coefficient includes a residual coefficient and Transformation coefficient.
The point cloud attribute encoding method according to claim 1, characterized in that the difference distance and the threshold distance are compared, and the current quantization parameter is adjusted according to the comparison result to obtain the current adjusted quantization parameter, or different ones are selected according to the comparison result. Rounding interval, the steps to obtain the current adjustment quantization rounding method include:

Compare the difference distance with the threshold distance;

If the difference distance is greater than the threshold distance, add the initial quantization parameter to the offset value to obtain the current adjusted quantization parameter, or select the first rounding interval as the current adjusted quantization rounding method;

If the difference distance is not greater than the threshold distance, select the initial quantization parameter as The currently adjusted quantization coefficient, or the second rounding interval is selected as the currently adjusted quantization rounding method.
The point cloud attribute encoding method according to claim 1, characterized in that, after the input coefficient is quantized using the currently adjusted quantization parameter or the currently adjusted quantization rounding method to obtain the quantized coefficient, the quantized coefficient is The steps to perform entropy coding and obtain the code stream include:

Predict the attribute value of the current point to be encoded to obtain a predicted value, calculate the residual coefficient between the predicted value and the real attribute value of the current point to be encoded; use the currently adjusted quantization parameter or the currently adjusted quantization rounding method to The residual quantized coefficients obtained by quantizing the residual coefficients are entropy encoded to obtain the code stream; wherein the input coefficients include the residual coefficients, and the quantized coefficients include the The residual quantization coefficient;

Or, predict the attribute value of the current point to be encoded to obtain the predicted value, calculate the residual coefficient between the predicted value and the real attribute value of the current point to be encoded, and transform the residual coefficient to obtain the transformation coefficient ; Using the transform quantization coefficient obtained by quantizing the transform coefficient using the currently adjusted quantization parameter or the currently adjusted quantization rounding method, performing entropy coding on the transform quantization coefficient to obtain the code stream; wherein, the The input coefficients include the transform coefficients, and the quantized coefficients include the transform quantized coefficients.
A computer-readable storage medium, characterized in that a point cloud attribute encoding program is stored on the computer-readable storage medium. When the point cloud attribute encoding program is executed by a processor, the point cloud attribute encoding program implements the requirements of any one of claims 1-3. The steps of the point cloud attribute encoding method described above.
A method for decoding point cloud attributes, characterized in that the method for decoding point cloud attributes includes the following steps:

Sort based on the point cloud data to be decoded according to preset rules to obtain the second sorted point cloud data; wherein the point cloud data to be decoded is point cloud data whose attributes are to be decoded;

Starting from the Nth point in the second sorted point cloud data, calculate the difference distance between the current point to be decoded and the most adjacent point according to the preset distance calculation method;

Compare the difference distance and the threshold distance, adjust the current inverse quantization parameter according to the comparison result, and obtain the current adjusted inverse quantization parameter;

Entropy decoding is performed on the encoded code stream to obtain reconstructed quantization coefficients, and the reconstructed quantization coefficients are inversely quantized according to the currently adjusted inverse quantization parameter to obtain reconstructed input coefficients.
The point cloud attribute decoding method according to claim 5, wherein the step of sorting the point cloud data to be decoded according to preset rules and obtaining the second sorted point cloud data specifically includes:

Obtain the target sorting code corresponding to the point cloud data to be decoded according to the original serial number arrangement rules, Morton order calculation rules or Hilbert order calculation rules; wherein the target sorting code is the input serial number or Morton code or Hill Bert code;

Sort the points in the point cloud data to be decoded in order from small to large or from large to small according to the target sorting code to obtain the second sorted point cloud data.
The point cloud attribute decoding method according to claim 5, characterized in that starting from the Nth point in the second sorted point cloud data, the current to-be-decoded point and the most similar point are calculated according to a preset distance calculation method. The steps for calculating the difference distance between adjacent points include:

According to the geometric distance calculation method, or the comprehensive distance calculation method composed of geometric distance and attribute distance, search for the point closest to the current point to be decoded as the nearest neighbor point of the current point to be decoded;

Or, use the first m points of the current point to be decoded as the most adjacent points of the current point to be decoded;

The distance difference between the current point to be decoded and the most adjacent point is calculated as the difference distance.
The point cloud attribute decoding method according to claim 5, wherein the step of comparing the difference distance and the threshold distance, adjusting the current inverse quantization parameter according to the comparison result, and obtaining the current adjusted inverse quantization parameter includes:

Compare the difference distance with the threshold distance;

If the difference distance is greater than the threshold distance, add the offset value to the initial inverse quantization parameter to obtain the current adjusted inverse quantization parameter;

If the difference distance is not greater than the threshold distance, the initial inverse quantization parameter is selected as the current adjusted inverse quantization parameter.
The point cloud attribute decoding method according to claim 5, characterized in that: performing entropy decoding on the encoded code stream to obtain reconstructed quantization coefficients, and performing inverse quantization on the reconstructed quantization coefficients according to the currently adjusted inverse quantization parameters. , the steps to obtain the reconstruction input coefficients include:

Perform entropy decoding on the encoded code stream to obtain transform reconstruction quantization coefficients or residual reconstruction quantization coefficients;

Using the currently adjusted inverse quantization parameter to perform inverse quantization on the transform reconstruction quantization coefficient or the residual reconstruction quantization coefficient, correspondingly obtain the transform reconstruction coefficient or the residual reconstruction coefficient; wherein the reconstruction input coefficient includes the transform reconstruction coefficient reconstruction coefficients and the residual reconstruction coefficients.
A computer-readable storage medium, characterized in that a point cloud attribute decoding program is stored on the computer-readable storage medium. When the point cloud attribute decoding program is executed by a processor, the point cloud attribute decoding program implements the requirements of any one of claims 5-9. The steps of the point cloud attribute decoding method described above.