Research On Inter-Layer Scalable Codec Algorithm

With the development of remote sensing technology,a large number of high-precision remote sensing images are produced.It has become an urgent problem that how to compress these data effectively.On the other hand,in the process of data communication,the transmission capacity of the channel and the request of the receiver to the image resolution are different,the user may expect high-precision images through a high-speed server,orobtain low-precision images through a low-speed modem,it has also become a major issue that how to satisfy the different requirements for transmission channels and images.SVC(Scalable Video Coding)is producedin order to solve the two major problems mentioned above,it can also be calledlayered video coding,the original image is compressed into bitstream at one time,and the images of different resolution are decoded according to different needs at the decoder.It is suitable for different occasions,the scalable coding and decoding scheme is the key point of this research.Until now,the pyramid model is usually used to implement scalable coding.The implementation of traditional pyramid model is relatively simple.The original image is used as the bottom layer,the second layer of image is obtained by downsamplingon the basis of the bottom layer,and the down sampling is continued on the basis of the second layer image to get the third layer imageand so on,the entire pyramid model is built.Finally,each layer is compressed and encoded to get the bit stream of each layer.In the traditional pyramid model,the correlation between layer and layer is not taken into account,therefore,the compression ratio after encoding is low.At the same time,the fifth-order filter is used in the down sampling process,and the loss of the low-frequency signal is high,so the noise-signal ratio is low after decoding.This paper proposes an inter-layer scalable codec algorithm that maximizes the image compression ratio while increasing the encoding quality.The main work and innovations of this paper are as follows:First,this paper has deeply researched pyramid model and compression coding algorithm,proposed an inter-layer scalable codec algorithm,and redesigned the pyramid framework of image,each layer only compresses the residual data of the original layer and the predicted layer to reduce the inter layer redundancy to the maximum extent.Second,according to the data characteristics of the new pyramid framework,the decimation filter and the interpolation filter are redesigned to hold more low-frequency components by reducing the order of filter.Third,on the basis of the traditional compression coding algorithm,the part of quantification which is the main part of the data loss is optimized and improved.A quantization based on compression ratio is proposed,the user is changed from giving a quantitative step to the expected compression rate directly,so the quantitative results are completely controlled by the user.On the other hand,a nonuniform quantization method is proposed in combination with the characteristics of quantized coefficients before quantizing.This scheme uses different quantization steps for different positions of each image block to retainmore low-frequency components of the image.Finally,the entropy coding of traditional compression coding algorithm is optimized and improved.Adaptive code table entropy coding is proposed,it has changed from the original 7 fixed code tables to generate the most suitable code tables adaptively according to the coefficient distribution after quantization,which greatly improves the compression ratio after coding and the image quality after decoding.