The problem of increasing the coding and decoding speed of Reed-Solomon codes (RS-codes) is considered. The implementation of the Open JPEG encoding system library was chosen as a base implementation for comparison. This task is considered for an x86-64 environment. The paper proposes ways to improve performance both when using a standard set of instructions and when using vector instructions from the SSSE3 and AVX2 sets. A technique for comparing the speed of encoding / decoding RS-codes is described using the example of the developed library and the basic implementation in Open JPEG. An experimental study of the relationship between the parameters of RS-codes, the encoding / decoding rate, and the set of instructions used is carried out. It is shown that for any RS-code it is possible to significantly increase performance even on a standard set of instructions. A method is proposed for the dynamic selection of the encoding acceleration method depending on the instructions supported by the target computer and the parameters of the used RS-code.

Keywords: anti-noize coding, Reed-Solomon codes, speed, vector instructions

The directions of modern research in the field of adaptive error correction, which can be applied when transmitting images in JPEG 2000 format over a noisy channel, are examined. Adaptive correction at the network level allows you to maintain a given percentage of packet delivery. Uneven protection at the application level is intended for use in broadcast broadcasts and allows the consumer to receive an image of various quality depending on the state of the network. Adaptive uniform protection at the application level is focused on point-to-point transmission and is aimed at receiving the maximum image quality by the receiver. The network layer tools discussed can be combined with the application layer tools.

Keywords: adaptation, noise-proof coding, forward error correction, noisy channel, JPEG 2000

The problem of transmitting an image in a transmitter-receiver system over a noisy data transmission channel in JPEG 2000 and JPWL format under conditions of a previously unknown and varying intensity of noise is considered. This task is considered for the conditions of the IP network. The article aims to build methods for estimating the intensity of noise by the JPWL decoding results in the receiver and dynamically adapting the parameters of the JPWL transmitter to the current assessment of the noise intensity, ensuring the highest quality of the decoded image with the shortest length of the transmitted encoded stream. An experimental study of the relationship between the parameters of JPWL, the intensity of noise and the quality of the decoded image was carried out. It is shown that for any JPWL parameter there is a region of one-to-one correspondence between the percentage of fully recovered tiles and the probability of losing a network packet. A technique for estimating the intensity of noise by the decoding results of a single image is described. This technique can be expanded to use the results of decoding a series of images.

Keywords: noisy channel, antinoise coding, noise intensity, decoding result, JPEG 2000, JPWL, experimental model

This article discusses image transfer in JPEG2000 format using JPWL antinoise coding via noisy channel with occurrence of packet errors. The main problem is recovery of received image with minimum deviations from input original. The aim of this work is to study experimentally the capability of JPWL protection to maintain quality of received and decoded image depending on the level of packet loss in data transmission channel. The experimental procedure consists of simulation of multiple transfer of JPWL protected image with the size of 1024×768 pixels and packet loss with the coded image data. Variable parameters of the study are: JPEG2000 coding procedure (irreversible conversion), Reed–Solomon codes used for JPWL protection, per cent of packet loss. Regular variant of JPWL protection is considered as well as combination of regular variant with algorithm of in-frame interleaving. The final experimental results are PSNR similarity extent of received and decoded image and input image, average per cent of completely recovered tiles with respect to tile number in code stream. The article describes the software complex developed for these studies, which includes JPEG2000 coder and decoder, JPWL coder and decoder, means of code stream dividing into RTP packets, simulating means of packet loss, assembling of frames from RTP packets. Operation flowcharts of JPEG2000 coder and decoder are described. It is mentioned that under conditions of packet errors JPEG2000 decoder should be able to process partially damaged or absent tiles. In-frame interleaving of code stream is described aimed at improvement of its stability with regard to packet errors. Three hypotheses are formulated. Hypothesis 1: in the case of packet loss the regular JPWL extension is unable to maintain quality of received and decoded image. Hypothesis 2: in the case of packet loss the regular JPWL extension in combination with algorithm of in-frame interleaving maintains the quality of received and decoded image. Hypothesis 3: in the case of packet loss and with application of algorithm of in-frame interleaving the PSNR value of received and decoded image depends statistically on the number of tiles completely recovered by JPWL decoder. As a result of the experimental study, all the hypotheses were fully confirmed.

Keywords: JPEG 2000, JPWL, packet errors, noise-immune encoding, decoded image quality