A Lossy Image Compression Method Using Autoencoder Neural Network

An image compression and neural network technology, applied in the field of image processing, can solve the problem of low compression rate, achieve the effect of improving compression rate and image quality

CN106254879BActive Publication Date: 2019-11-08GUANGDONG KINGPOINT DATA SCI & TECH CO LTD
6 Cites 0 Cited by

Patent Information

Authority / Receiving Office
CN · China
Patent Type
Patents(China)
Current Assignee / Owner
Publication Date
2019-11-08

Smart Images

  • Figure 1
    Figure 1
  • Figure 2
    Figure 2
  • Figure 3
    Figure 3
Patent Text Reader

Abstract

The invention relates to a lossy image compression method adopting a self-encoding neural network. The method includes the following steps that: S1, a lossy image is preprocessed, so that a sampled image can be obtained; S2, a self-encoding neural network model is established; S3, a hidden layer image is calculated according to the self-encoding neural network model; and S4, the hidden layer image is adopted as a new sampled image, and subsequent compression processing is performed on the new sampled image, so that a final compressed image. Compared with the prior art, the lossy image compression method of the present invention uses the self-encoding neural network in mapper constructing process, so that redundant information processing, namely, image dimensionality reduction, is performed on the image again; and at the same time, the network has a function similar to feature extraction, which means that a certain denoising effect can be achieved in an image decompression and restoring process by using the effects of the features of the hidden layer, and therefore, the quality of the image can be enhanced with a compression ratio improved.
Need to check novelty before this filing date? Find Prior Art

Description

technical field

[0001] The invention relates to the technical field of image processing, in particular to a lossy image compression method using a self-encoding neural network. Background technique

[0002] With the advent of the era of big data, data is growing at an explosive rate, and huge amounts of data carry information between people. As the visual basis for human perception of the world, images are the basis for human beings to obtain information, express information and transmit information. important means. Therefore, how to ensure the accurate and fast transmission of images has become one of the important issues in digital image processing. The most direct method is to compress the image, that is, to reduce the amount of data required to represent the digital image, so that the transmission rate of the image can be increased while also The integrity and accuracy of image information can be guaranteed. Image compression is an important key technology of image pr...

Examples

Embodiment Construction

[0036] The above and other technical features and advantages of the present invention will be described in more detail below in conjunction with the accompanying drawings.

[0037] like figure 1 Shown is a flow chart of a lossy image compression method using self-encoding neural network of the present invention, the method may further comprise the steps:

[0038] Step S1: Preprocessing the lossy image to obtain a sampled image.

[0039] Specifically, the preprocessing includes: performing color mode conversion on the original lossy image, and determining an image sampling method as required to sample the converted image to obtain a sampled image of the corresponding image. The color mode conversion is specifically to convert the RGB color space into the YCrCb color space. Two sampling methods are usually used: YUV411 and YUV422. Correspondingly, it means that the data sampling ratio of the three components of Y, Cb, and Cr is generally 4:1:1 or 4:2:2.

[0040] Step S2: Est...