CN108200433B

CN108200433B - Image compression and decompression method

Info

Publication number: CN108200433B
Application number: CN201810109934.7A
Authority: CN
Inventors: 张辉
Original assignee: Chongqing Lechi Information Technology Co Ltd
Current assignee: Chongqing Lechi Information Technology Co., Ltd
Priority date: 2018-02-05
Filing date: 2018-02-05
Publication date: 2020-01-07
Anticipated expiration: 2038-02-05
Also published as: CN108200433A

Abstract

The invention discloses an image compression and decompression method, which comprises the steps of dividing a plurality of images into a plurality of sub-regions to produce reduced images, and constructing a decompression table according to the pixel distribution condition of the sub-regions and the sub-regions adjacent to the sub-regions for later image restoration. The decompression table constructed according to the embodiment of the invention can help to further reduce the storage space and processing time required for decompressing data when a plurality of content-similar pictures are processed in batch, and reduce the visual impact of the quality degradation caused by image compression on users.

Description

Image compression and decompression method

Technical Field

The present invention relates to the field of image processing technology, and more particularly to a method for compressing and decompressing an image.

Background

In image processing and storing, the size of the picture often needs to be compressed in order to save processing resources and storage space. For example, when the original picture may have too many pixels and the original picture is reduced in size in actual use without being visually affected, the original picture may be reduced to one fourth, and then the picture compression technique is used to further reduce the size of the image, so that the processing time and the storage space may be reduced to about one fourth. In application scenarios where images are processed in large volumes, similar operations are very frequent and necessary. The user often needs to upload a large number of high-resolution pictures shot by the camera to the mobile phone and modify the pictures through the APP on the mobile phone, and because the performance of a processor and the storage space of the mobile phone are large relative to the difference between the processor and the PC, the pictures generally need to be reduced and compressed before image modification operations such as a filter are performed. In addition, the mobile phone user may need to perform similar operations on multiple similar images, such as performing batch rapid image modification, or performing dynamic custom expression creation. Because the performance of the mobile phone image modification APP is very limited compared with a professional image modification system at a PC end, and the accuracy of the user touch screen operation is greatly reduced compared with that at the PC end, a large amount of image reduction and compression needs to be performed in an automatic image processing manner so as to adapt to operations such as image modification, dynamic expression production, short video splicing or simple slide editing and the like performed on a mobile phone. Picture compression is currently mainly divided into two broad categories, lossy compression, which typically includes RAW format images taken directly by a camera, and lossless compression, which includes popular JPEG lossy compression format images, for example. Lossless compression may be lower than lossy compression, but may better preserve image quality, and lossy compression, although the compression ratio is more desirable, typically discards some kinds of image information in the image, and may cause more noticeable image artifacts when printed or enlarged. How to improve the quality of the compressed image as much as possible while ensuring the processing speed is the key to provide better user experience. Especially, the image quality is more easily damaged by image compression after the image is reduced on the mobile phone, so a simple and easy method is needed to make a compromise between the compression rate and the image quality according to the images with different properties. The image processed by the image modification APP on the mobile phone may include different types of scenes, for example, when a scene with strong contrast such as characters and a human face is shot, the image quality is damaged and then is more easily perceived by a user after being reduced, and when a scene with a complex background such as a natural landscape and a performance field is shot, the image quality is not easily perceived by the user even if the image quality is damaged, so that a compression method and a reduction method should be selected in a targeted manner through judgment of image content, so that processing resources and storage space required particularly when a plurality of pictures are modified in batch on the premise that the user is not aware of changes in vision are saved as much as possible, and the user can obtain pictures or videos suitable for being reproduced on the mobile phone through a more convenient processing method.

Disclosure of Invention

An object of the present invention is to provide an image encoding method for saving processing efficiency and storage space in batch image compression or decompression.

Embodiments of the present invention relate to an image compression and decompression method including spatially and identically dividing a plurality of images into a plurality of square sub-regions of identical size, each sub-region including four pixels respectively located in first, second, third and fourth quadrants in a cartesian coordinate system; taking the pixel value in the first quadrant of each sub-area of each image as the pixel value corresponding to the sub-area in the reduced image to obtain a reduced image with the image size reduced to one fourth; determining each subregion of each image, and respectively determining four first-type subregions with equal pixel values, four second-type subregions which contain two pixel values and have the same pixel value as another pixel value except the first quadrant pixel value of the subregion in the nearest neighbor quadrant, four third-type subregions which contain two pixel values and have no pixel value as another pixel value except the first quadrant pixel value of the subregion in the nearest neighbor quadrant, and four fourth-type subregions which contain more than three pixel values, wherein the nearest neighbor quadrant is an adjacent quadrant in adjacent subregions of each of the second, third and fourth quadrants; and creating a decompression table for each image, the decompression table containing decompression data for pixels of the second, third and fourth quadrants corresponding to each sub-region and the number of sub-regions of the first, second, third and fourth types, the decompression data comprising for a type of sub-region only one of a pixel value and a nearest neighbor quadrant position having the same pixel value.

In some embodiments, each reduced image is stored in association with a corresponding decompression table.

In some embodiments, each reduced image is stored in association with an offset of the corresponding decompression table, where the offset is a difference between the decompressed data of the decompression table corresponding to one reduced image and the decompressed data of the decompression table corresponding to the previous reduced image.

In some embodiments, the decompressed data for the sub-region of the second type includes a pixel distribution layout within the sub-region and a position of a nearest neighbor quadrant having the same pixel value when the nearest neighbor quadrant has the same pixel value.

In some embodiments, the decompressed data of the third-type sub-region and the fourth-type sub-region comprises pixel values of the second, third and fourth quadrants.

In some embodiments, the reduced image is losslessly compressed when the sum of the number of first-type and second-type sub-regions stored in the decompression table is greater than the sum of the number of third-type and fourth-type sub-regions.

In some embodiments, the reduced image is lossy compressed when the sum of the number of first-type and second-type sub-regions stored in the decompression table is less than or equal to the sum of the number of third-type and fourth-type sub-regions.

In some embodiments, the plurality of images are losslessly compressed when the sum of the number of all of the first-type and second-type sub-regions is greater than the sum of the number of all of the third-type and fourth-type sub-regions in the decompression table for the plurality of images.

In some embodiments, the plurality of images are lossy compressed when the sum of the number of all of the first-type and second-type sub-regions in the decompression table of the plurality of images is less than or equal to the sum of the number of all of the third-type and fourth-type sub-regions.

In some embodiments, the pixel value is any one of an RGB color channel value and a transparent channel value.

Embodiments of the present invention advantageously include batch processing of multiple pictures on a cell phone with greater processing resource and storage space utilization efficiency, while ensuring that the compressed image quality degradation is not visually perceptible to the cell phone user. The embodiment of the invention also allows the compressed picture to be restored to the original picture quality at any time through the decompression table, so that the user can conveniently process the pictures on the mobile phone in batch for the mobile phone to use in a more automatic mode, and the original picture can be restored on the PC section later under the condition of saving the storage space so as to carry out more detailed picture modification. A decompression table constructed according to an embodiment of the present invention will help to further reduce the memory space required for decompressing data in the case where, for example, pictures for batch processing such as manufacturing dynamic expressions are similar.

Drawings

The accompanying drawings are provided to illustrate embodiments in conjunction with the description, but are not intended to be limiting.

FIG. 1 is a schematic illustration of image sub-region partitioning according to some embodiments.

FIG. 2 is a flow diagram of a method of compressing an image according to some embodiments.

Detailed Description

It will be understood by those skilled in the art that, although the terms first, second, etc. may be used herein to describe various elements, these elements should not be limited by these terms. These terms are only used to distinguish one element from another. For example, a first element could be termed a second element, and, similarly, a second element could be termed a first element, without departing from the scope of the present invention. As used herein, the term "and/or" includes any and all combinations of one or more of the associated listed items.

FIG. 1 is a schematic illustration of image sub-region partitioning according to some embodiments. The plurality of equally sized images will be spatially divided in the same way into a plurality of equally sized square sub-regions, each sub-region comprising four pixels located in a first, second, third and fourth quadrant, respectively, of a cartesian coordinate system. Thus, for example, a picture of 1024 × 768 would be divided into 196608 sub-regions and could be mapped to a small size picture of 512 × 384. As shown in fig. 1, the first sub-area includes four

pixels

101, 102, 103, and 104 located in the first, second, third, and fourth quadrants, respectively. Similarly, the second sub-area includes four

pixels

111, 112, 113, and 114 respectively located in the first, second, third, and fourth quadrants, the third sub-area includes four

pixels

121, 122, 123, and 124 respectively located in the first, second, third, and fourth quadrants, and the fourth sub-area includes four

pixels

131, 132, 133, and 134 respectively located in the first, second, third, and fourth quadrants. Except for the edge quadrant of a sub-area located at the edge or four corners of the image, the remaining quadrants will have nearest neighbor quadrants, i.e. adjacent quadrants in the adjacent sub-area of that quadrant. For example, the fourth quadrant 104 in the first sub-region has three

nearest neighbor quadrants

113, 121, and 132, and similarly, the first quadrant 121 in the third sub-region also has three

nearest neighbor quadrants

104, 113, and 132.

FIG. 2 is a flow diagram of a method of compressing an image according to some embodiments. In step S201, after a plurality of images are read, each image is divided into sub-regions according to the above-described division method. After the sub-region division is completed, a pixel value in one quadrant of each sub-region of each image is mapped to one pixel in the reduced image to obtain a reduced image in which the image size is reduced to one quarter. This is mainly to reduce the processing time required for the image reduction operation, and although other sampling methods such as averaging, interpolation, etc. may be used to determine the pixels in the reduced image, it is more efficient to directly use the pixel values in one quadrant as representative pixels, and it is sufficient to provide the required image quality for the moving end. The quadrant used as the representative pixel may be any one of the first, second, third and fourth quadrants, and the first quadrant is used herein as the representative pixel by way of example.

In step S202, the distribution of pixels for each sub-region is determined to reduce the processing and memory resources required for decompressing data for different types of sub-regions. Herein, the pixel value of each pixel of the sub-region may be any one of an RGB color channel value and a transparent channel value. For convenience of the following description, each pixel value contains, for example, 8 bits. The RGB color channel and the clear channel may be processed separately or together when the decompression process is performed. If the pixel values of all pixels in the sub-region are consistent, for example, the four

pixels

101, 102, 103, and 104 of the first sub-region in fig. 1 all have the same pixel value, then the decompressed data does not need to be produced at this time, and the whole sub-region can be directly restored from the pixel 101. The sub-regions where the four pixel values are opposite are referred to as first type sub-regions. Furthermore, a sub-region of the second type should be determined which contains both pixel values and which also includes in the nearest neighbor quadrant the same pixel value as another pixel value than the first quadrant pixel value of the sub-region. For example, the representative pixel in the second sub-area comprising four

pixels

111, 112, 113 and 114 in fig. 1 is 111, whereas the

pixels

112 and 114 are identical to 111, only the pixel 113 has a different color, but the pixel value of the pixel 121 in the third sub-area is identical to 113. At this point, instead of using 8 bits to store the pixel value of the pixel 113, 2 bits of position data may be used to store the nearest quadrant position having the same pixel value, such as the nearest pixel 121 with 10 to indicate 113. When two or more of the three nearest neighbor pixels have the same pixel value, the position of any one of them may be stored. To further reduce the decompressed data size of a sub-area containing two colors, the pixel distribution layout within the sub-area may be defined in 3 bits, e.g. 111 indicates that the pixels outside the first quadrant coincide, 001 indicates that the second and third quadrants are different from the pixels of the first quadrant and the fourth quadrant is the same as the first quadrant, 101 indicates that the second and fourth quadrants are the same as the first quadrant and the third pixel is different from the first quadrant, etc. In the case of including two kinds of pixels, the pixel distribution layout is 8 kinds in total, and the correspondence with the above 3 bits can be made by a simple correspondence table. A third type sub-region should also be determined which contains both pixel values and which does not include in the nearest neighbor quadrant the same pixel value as another pixel value than the first quadrant pixel value of the sub-region, in which case the pixel value different from the representative pixel cannot be determined by 2 bits of position data, requiring the pixel value to be stored in decompressed data. Both of these methods may be more efficient than storing 8 bits of decompressed data. Finally, a fourth type of sub-area comprising three or more colors should also be determined, for which case the resources saved by using the pixel distribution layout and nearest neighbor quadrants to store decompressed data are more limited than for directly storing pixels, and therefore pixel values can be stored directly. In summary, the decompressed data of the second type sub-region comprises the pixel distribution layout within the sub-region and the position of the nearest neighbor quadrant having the same pixel value when having the same pixel value in the nearest neighbor quadrant. And the decompressed data of the third-type sub-region and the fourth-type sub-region comprise pixel values of the second, third and fourth quadrants. For a general picture, the second type and the number of sub-regions occupy most of the proportion when the sub-regions include four pixels, and the key to improve the resource utilization efficiency is to process the sub-regions of the second type.

In step S203, a decompression table of the image is built according to the above steps, the decompression table containing the decompressed data of the pixels of the second, third and fourth quadrants corresponding to each sub-region and the number of the first, second, third and fourth type sub-regions. For sub-regions of the first type, no decompressed data need be provided, and only a 1-bit indicator bit may be used to indicate that one pixel in the reduced image is used to revert to the original sub-region. For a sub-region of the second type, the above-mentioned pixel distribution situation within the sub-region and the nearest neighbor quadrant situation with the same pixel value are stored into a decompression table without storing the pixel value. The opposite is true for sub-regions of the third and fourth types, storing pixel values in the compression table without storing pixel distribution conditions or nearest neighbor quadrant conditions. Each reduced image may be stored in association with a corresponding decompression table on the same or a different memory for use in restoring the picture at any time. In the case of processing a plurality of images in batch, each reduced image is stored in association with the offset of the corresponding decompression table, the offset is the difference between the decompressed data of the decompression table corresponding to one reduced image and the decompressed data of the decompression table corresponding to the previous reduced image, and may only include the position where the change occurs, which is advantageous for processing the custom expression or the motion picture with small difference between adjacent images. The decompression table further includes the number of sub-regions of the first type, the second type, the third type, and the fourth type. These numbers can be used to determine which compression method to use for the reduced image to further improve storage efficiency. The reduced image may be losslessly compressed when the sum of the number of first-type and second-type sub-regions stored in the decompression table is greater than the sum of the number of third-type and fourth-type sub-regions. At this time, the picture is generally relatively simple, and may include objects with obvious edges, such as characters, human faces, and the like, and lossless compression is required to avoid the occurrence of obvious visible picture quality degradation, otherwise lossy compression is performed on the basis of picture reduction, so that the situations of obvious fracture, noise, blur, and the like easily occur when a user amplifies or decompresses the picture. And lossy compressing the reduced image if the sum of the numbers of the first-type and second-type sub-regions stored in the decompression table is less than or equal to the sum of the numbers of the third-type and fourth-type sub-regions. In this case, since pictures generally include relatively complicated image information such as paintings, CG, nature scenes, and the like, and are difficult to visually find by a user even when lossy compression is performed, the compression ratio can be increased by the lossy compression. The relative ratio of the number of sub-regions of the first and second type to the number of sub-regions of the third and fourth type is a relatively simple measure of how lossless or lossy compression is performed. On the basis of the above, other indexes such as the ratio of the number of the first-type subareas to the number of the fourth-type subareas, the ratio of the number of the second-type subareas to the number of the third-type subareas, and the like, which are easily thought by those skilled in the art, can be judged. When a plurality of images such as moving pictures, expressions and the like are processed, if the sum of the number of all the sub-areas of the first type and the second type in the decompression table of the plurality of images is larger than the sum of the number of all the sub-areas of the third type and the fourth type, the plurality of images are compressed in a lossless mode. And when the sum of the number of all the first-type and second-type sub-regions in the decompression table of the plurality of images is less than or equal to the sum of the number of all the third-type and fourth-type sub-regions, performing lossy compression on the plurality of images. Thus, when a plurality of images are processed, the compression mode can be selected according to the overall characteristics of the images.

The embodiments described above are merely illustrative of the principles of the present invention and alternate or equivalent embodiments that may be envisioned by those skilled in the art upon viewing the figures and description presented herein are intended to be within the scope of the present invention.

Claims

1. An image compression and decompression method, characterized by comprising:

dividing a plurality of images into a plurality of square sub-areas with the same size in a same way in space, wherein each sub-area comprises four pixels respectively positioned in a first quadrant, a second quadrant, a third quadrant and a fourth quadrant in a Cartesian coordinate system;

taking the pixel value in the first quadrant of each sub-area of each image as the pixel value corresponding to the sub-area in the reduced image to obtain a reduced image with the image size reduced to one fourth;

determining each subregion of each image, and respectively determining four first-type subregions with equal pixel values, a second-type subregion containing two pixel values and a nearest neighbor quadrant which does not contain the same pixel value as another pixel value except the first quadrant pixel value of the subregion, a third-type subregion containing two pixel values and a nearest neighbor quadrant which does not contain the same pixel value as another pixel value except the first quadrant pixel value of the subregion, and a fourth-type subregion containing more than three pixel values, wherein the nearest neighbor quadrant is an adjacent quadrant in adjacent subregions of each of the second, third and fourth quadrants; and creating a decompression table for each image, the decompression table containing decompression data for pixels of the second, third and fourth quadrants and the number of sub-regions of the first, second, third and fourth types corresponding to each sub-region, the decompression data comprising for a type of sub-region only one of a pixel value and a nearest neighbor quadrant location having the same pixel value; and

lossless compression is performed on the reduced image when the sum of the numbers of the first-type and second-type sub-regions stored in the decompression table is greater than the sum of the numbers of the third-type and fourth-type sub-regions, and lossy compression is performed on the reduced image when the sum of the numbers of the first-type and second-type sub-regions stored in the decompression table is less than or equal to the sum of the numbers of the third-type and fourth-type sub-regions.

2. The method of claim 1, further comprising storing each of the reduced images in association with the corresponding decompression table.

3. The method of claim 1, further comprising storing each of the reduced images in association with an offset of the corresponding decompression table, wherein the offset is a difference between the decompressed data of the decompression table corresponding to one of the reduced images and the decompressed data of the decompression table corresponding to the previous reduced image.

4. The method of claim 1, characterized in that the decompressed data of the sub-area of the second type comprises a pixel distribution layout within the sub-area and a position of a nearest quadrant having the same pixel value when having the same pixel value in the nearest quadrant.

5. The method of claim 4, wherein said decompressed data for said third-type sub-region and said fourth-type sub-region includes pixel values for said second, third, and fourth quadrants.

6. The method of claim 5, further comprising lossless compressing the plurality of images when a sum of a number of all of the first-type and second-type sub-regions in the decompression table for the plurality of images is greater than a sum of a number of all of the third-type and fourth-type sub-regions.

7. The method of claim 6, further comprising lossy compressing the plurality of images when a sum of a number of all of the first-type and second-type sub-regions in the decompression tables for the plurality of images is less than or equal to a sum of a number of all of the third-type and fourth-type sub-regions.

8. The method of claim 1, wherein the pixel value is any one of an RGB color channel value and a clear channel value.