Study Of Thermal Infrared Image Compression Based On Wavelet

With the development of infrared image capturing technology, its information capacity almost redoubled. As thermal image is difficult to be adopted in practical application due to its bandwidth resource consumption caused by huge data flow, thermal infrared image compression has become an important subject in industrial automatic detection.Firstly, this paper introduces the existing natural image compression standard and algorithm. Previous infrared image compression systems usually adopted algorithms based on mature natural image compression standard. Image compression algorithm has become a focus and hot topic in image processing research. Although there are many mature algorithms applied in practice, each of them has limitation for their special application environment. As thermal infrared image characteristic is different from natural images, general image compression algorithm may be not reasonable choice for thermal infrared image. So that we must research compression algorithm based on the analysis of thermal infrared image characteristic, which brings forward higher requirement to the study of thermal infrared image compression.Secondly, key technologies of image compression based on wave let transform, such as wavelet basis selection and wavelet image coefficients distribution, are introduced after the study of wavelet analysis theory. The aim is to research key technology of image compression considering existing image analysis standard, and put forward a more effective compression algorithm for industrial detection based on the infrared image characteristic.This paper is focused on image compression algorithm. After investigating embedded zero-tree wavelet coding (EZW) algorithm and multistage tree set division (SPIHT) coding algorithm, SPIHT is improved based on the characteristic of thermal infrared image and wavelet to achieve a more suitable compression algorithm for thermal infrared image.The simulation results show that the improved compression algorithm proposed in this paper can recover image when the image compression ratio is remained. With the increase of wavelet decomposition layers, PSNR value of this improved algorithm is significantly higher than EZW algorithm and SPIHT algorithm under the condition of the same compression ratio, which can illustrate its superiority. As the image is divided into main and subsidiary images and they are compressed respectively, the important areas are relatively more clear after recovery, which can improve the visual effect. At the same time, it can be seen from the experiment results, the average temperature almost remains the same, which means that deep compression ot non-important areas does not have big influence on temperature. PSNR is improved after processed using the proposed method, especially the subjective image quality is improved significantly.