Beta
166190

A Neural Network Based Hybrid Technique for Data Compression of Stereopairs.

Article

Last updated: 22 Jan 2023

Subjects

-

Tags

Electrical Engineering

Abstract

This paper presents a new hybrid technique for efficient compression of stereopairs . The proposed technique encompasses three processing stages: a phase difference disparity interpreter, a neural network based image compressor / decompressor, and a stereopair reconstructor. The main advantage of the proposed technique lies in a threefold data compression. The first data compression stage is realized by using the Fast Fourier Transform (FFT) for computation of the disparity. Data compression is achieved by sending both the disparity and the compressed right image. The second data compression stage is achieved by clamping the significant part of the FFT coefficients of the right image to the input layer of the neural network for further data compression. This has the advantage of reducing the number of neurons in the input layer of the neural compressor. The third data compression stage is encompassed in the data compression of the previously FFT - compressed right image through the neural net data compressor. At the receiver, the disparity, the displaced subimage and the compressed right image are used to reconstruct the left image: Experiments are performed on both random and real stereopairs. The compression ratio of the neural network depends on the number of neurons in the hidden layer. The FFT of the input parterns is used to increase the compression ratio and to speed up the learning of the neural network. Correspondence analysis is performed using the spin model to obtain the disparity map of the stereopair. System design parameters affecting the performance of the correspondence analyzer are studied to achieve optimal stereopais interpretation. This process makes it possible to transmit only the displaced regions in the left image. Another data compression step is achieved through the contour of the disparity map. 

DOI

10.21608/bfemu.2021.166190

Authors

First Name

Ahmad

Last Name

Tolba

MiddleName

-

Affiliation

Electrical Power Engineering Department., Faculty of Engineering., Suez Canal University., Port Said.

Email

-

City

-

Orcid

-

First Name

Abdul

Last Name

Ragab

MiddleName

H.

Affiliation

Electrical Power Engineering Department., Faculty of Engineering., Menoufia University., Menoufia.

Email

-

City

Menoufia

Orcid

-

Volume

18

Article Issue

4

Related Issue

24036

Issue Date

1993-12-01

Receive Date

1993-10-11

Publish Date

2021-12-01

Page Start

1

Page End

17

Print ISSN

1110-0923

Online ISSN

2735-4202

Link

https://bfemu.journals.ekb.eg/article_166190.html

Detail API

https://bfemu.journals.ekb.eg/service?article_code=166190

Order

4

Type

Research Studies

Type Code

1,205

Publication Type

Journal

Publication Title

MEJ. Mansoura Engineering Journal

Publication Link

https://bfemu.journals.ekb.eg/

MainTitle

-

Details

Type

Article

Created At

22 Jan 2023