CN112004094A - A method of saving pixel storage space - Google Patents

Abstract

The present application provides a method for saving pixel storage space, the method specifically includes the following steps: S1, a frame space is between the starting point of the first frame writing and reconstructing pixel and the first frame writing and reconstructing pixel end point; the first frame The end position of the write reconstruction pixel is the starting point of the configurable expansion space of the storage space; S2, the starting point of the second frame writing and reconstructing pixel is the starting point of the configurable expansion space of the first frame in step S2, and the second frame writes the reconstructed pixel to the storage space. After the end of the space, go back to the starting point of the space to write the following pixels; S3, repeat steps S1-S2 in the following frame and so on, jump up and expand the number of lines to start writing the reconstruction data, and write the reconstructed pixel starting point of the previous frame. Becomes the starting point of the read reconstructed pixel for the current frame.

Description

A method of saving pixel storage space

technical field

The invention relates to the technical field of image processing, in particular to a method for saving pixel storage space.

Background technique

Today's society is an information society, and there are more and more image materials. The storage space requirements for storing image material are also increasing, and during the application process, especially in video codecs, as shown in Figures 1 and 2, there is a need for constant data exchange with off-chip memory, including reconstructed frame writing input, reference frame acquisition, etc. This requires at least twice the pixel space of a frame for storage. That is, the image transmission also needs to occupy a certain storage space. If the storage space of images can be reduced, resources will be greatly saved and transmission efficiency will be improved.

In the prior art, the space of two frames is used to store the reconstructed frame and the reference frame respectively, which has strict requirements on the memory size. If other technologies such as multiple reference frames are added, the use of memory resources will be more intensive.

In the conventional method, as shown in FIG. 3 , the first frame and the second frame in FIG. 3 are used as examples. The encoded reconstructed frame pixels of the first frame are written into the reconstructed frame pixel storage space of the first frame. It is used for inter-frame prediction when encoding the second frame. At the same time, the second frame also has reconstructed pixels written into the second frame reconstructed frame pixel storage space during encoding. However, the reconstructed pixels in the first few lines of the first frame have no effect in the post-encoding of the second frame, and the inter-frame prediction will not search for a very far position, that is, there is storage redundancy.

SUMMARY OF THE INVENTION

In order to solve the problems existing in the above-mentioned prior art, the purpose of the present invention is to reduce memory occupation, and achieve shared use of memory without affecting coding, thereby saving storage space.

The present application provides a method for saving pixel storage space, wherein the storage space is set as a frame space and an expansion space with a configurable space size; a frame is between the starting point of the writing and reconstructing pixels of the current frame and the ending point of the writing and reconstructing pixels of the next frame. space; the end position of the write reconstruction pixel of the next frame is the start point of the configurable expansion space of the storage space.

The method specifically includes the following steps:

S1, the 1st frame write reconstruction pixel start point and the 1st frame write reconstruction pixel end point is a frame of space; Described 1st frame write reconstruction pixel end point position is the start point of the configurable expansion space of the storage space;

S2, the starting point of the 2nd frame writing and reconstructing pixel is the starting point of the configurable expansion space of the 1st frame in step S2, and after the 2nd frame writing and reconstructing the pixel to the end point of the storage space, it returns to the starting point of the space to write the following pixels;

S3, repeating steps S1-S2 in the same way for subsequent frames, jumping upward to expand the number of lines to start writing reconstruction data, and the starting point of the writing reconstruction pixel of the previous frame becomes the starting point of the reading reconstruction pixel of the current frame.

In the step S3, if the jump is smaller than the address of the starting point of the space, the remaining part will jump forward from the ending point of the space.

The configurable number of lines in the expansion space is a multiple of 64, and is not greater than the height of one frame.

The theoretical minimum value of the expansion space is 64 lines of pixels.

In the above steps, even if the reconstructed pixels of the next frame are written from the starting point of space and cover the reconstructed pixels of the previous frame, because of the constraints of the search range, the pixels of this starting point are useless reference pixels at this time, so they can be covered. of.

Description of drawings

The accompanying drawings described herein are used to provide a further understanding of the present invention, and constitute a part of the present application, and do not constitute a limitation to the present invention.

FIG. 1 is a schematic block diagram of data exchange in a video codec in the prior art.

FIG. 2 is a schematic block diagram of the storage space of the reconstructed pixel of the first frame and the storage space of the reconstructed pixel of the second frame in the prior art.

FIG. 3 is a schematic block diagram of the use of the first frame reconstructed pixel storage space and the second frame reconstructed pixel storage space in the prior art.

FIG. 4 is a schematic diagram of the new storage space and configurable expansion space of the present application.

FIG. 5 is a schematic diagram of the start and end positions of the reconstructed pixels in the first frame of the present application.

FIG. 6 is a schematic diagram of writing the starting point of the reconstructed pixel in the second frame of the present application and writing the following pixels after returning to the starting point of the space after the second frame writing the reconstructed pixel to the spatial end point.

FIG. 7 is a schematic diagram of writing the starting point of a reconstructed pixel in the second frame, the third frame and the fourth frame of the present application.

FIG. 8 is a schematic diagram of writing the starting point of the reconstructed pixel in the fourth frame, the fifth frame and the sixth frame of the present application.

Detailed ways

At present, the commonly used terms in image storage technology are:

Expansion space: The pixel space expanded based on the original memory space, in pixel row units.

Raw Frame: A frame in the raw video image.

Reconstructed frame: The reconstructed image after encoding and decoding the original frame.

Reference frame: When encoding the original frame, the frame used for reference for prediction, usually the encoded reconstructed frame.

Inter prediction: A prediction method that searches for the best matching pixel unit based on the reference frame.

As shown in Figure 4, memory optimization. Since there is redundancy in practical applications, we do not need twice the memory space, but only need to expand some pixel rows on the basis of a memory space, and the useless reference area can be written to reconstruct pixels. That is, a new combination of storage space and reference space is formed: one frame space and configurable expansion space.

Figure 5 shows how frame 1 writes the reconstructed frame. The starting point of the reconstruction pixel is written in the first frame, and the end point of the reconstructed pixel is written in the first frame; the position of the end point of the writing reconstruction pixel is the starting point of the new configurable expansion space of the storage space.

As shown in Figure 6, it shows how to write the reconstructed frame and read the reference frame when the second frame is performed. The starting point of writing and reconstructing pixels of the second frame is the starting point of the configurable expansion space of the original first frame. After writing the reconstructed pixels of the second frame to the ending point of the original storage space, it returns to the starting point of the space to write the following pixels.

Among them, the configurable expansion space is directly related to the search range of inter-frame prediction, so it cannot be very small. The theoretical minimum value is 64 lines of pixels, and the number of lines in the configurable expansion space is a multiple of 64, not greater than the height of one frame. .

In particular, there is a pixel coverage problem here, even if the reconstructed pixels of the second frame are written from the spatial starting point and cover the reconstructed pixels of the first frame, because of the constraints of the search range, the pixels at this starting point are useless reference pixels at this time. So it can be overwritten.

The subsequent frames jump upward in turn to expand the number of lines to start writing reconstruction data, and the starting point of the writing reconstruction pixel of the previous frame becomes the starting point of the reading reconstruction pixel of the current frame. If the jump is smaller than the start address of the space, the remaining part will jump forward from the end point of the space.

Specifically, as shown in FIG. 7 , the starting point of the writing and reconstructing pixels of the second frame is shown to jump upward to expand the number of lines to start writing the reconstructed data, and the starting point of the writing and reconstructing pixels of the first frame becomes the starting point of the reading and reconstructing pixels of the current frame; and so on, the third The starting point of the frame write reconstruction pixel jumps upward to expand the number of lines to start writing the reconstructed data, the starting point of the second frame write reconstruction pixel becomes the starting point of the read reconstruction pixel of the current frame; the starting point of the fourth frame write reconstruction pixel jumps upward to expand the number of lines to start writing the reconstructed data , the starting point of the write reconstruction pixel of the 4th frame becomes the starting point of the read reconstruction pixel of the current frame; as shown in Figure 8, if the jump is smaller than the address of the space start point, the remaining part will jump forward from the space end point, that is, the 5th frame jumps If it is smaller than the address of the starting point of the space, the starting point of the write reconstruction pixel in the sixth frame jumps forward from the ending point of the space.

The above descriptions are only preferred embodiments of the present invention, and are not intended to limit the present invention. For those skilled in the art, various modifications and changes may be made to the embodiments of the present invention. Any modification, equivalent replacement, improvement, etc. made within the spirit and principle of the present invention shall be included within the protection scope of the present invention.

Claims (6)

1. a saving method of pixel storage space, it is characterized in that, described storage space is set as the expansion space of a frame space and configurable space size; The current frame writes the reconstruction pixel starting point and the next frame writes the reconstruction pixel end point between. One frame of space; the end position of the write reconstruction pixel of the next frame is the start point of the configurable expansion space of the storage space. 2. The method for saving a pixel storage space according to claim 1, wherein the method specifically comprises the following steps: S1, the 1st frame write reconstruction pixel start point and the 1st frame write reconstruction pixel end point is a frame of space; Described 1st frame write reconstruction pixel end point position is the start point of the configurable expansion space of the storage space; S2, the starting point of the 2nd frame writing and reconstructing pixel is the starting point of the configurable expansion space of the 1st frame in step S2, and after the 2nd frame writing and reconstructing the pixel to the end point of the storage space, it returns to the starting point of the space to write the following pixels; S3, repeating steps S1-S2 in the same way for subsequent frames, jumping upward to expand the number of lines to start writing reconstruction data, and the starting point of the writing reconstruction pixel of the previous frame becomes the starting point of the reading reconstruction pixel of the current frame. 3 . The method for saving pixel storage space according to claim 1 , wherein, in the step S3, if the jump is smaller than the starting point address of the space, the remaining part jumps forward from the end point of the space. 4 . 4 . The method for saving pixel storage space according to claim 1 , wherein the configurable number of rows in the expansion space is a multiple of 64, and is not greater than the height of one frame. 5 . 5 . The method for saving pixel storage space according to claim 1 , wherein the theoretical minimum value of the expansion space is 64 rows of pixels. 6 . 6. a kind of saving method of pixel storage space according to claim 1, is characterized in that, even if the reconstruction pixel of the following frame starts to write from the space starting point in the described step, covers the reconstruction pixel of the previous frame, because the search Constraints on the range, at this point the pixel at this starting point is already a useless reference pixel, so it can be covered.

Images