Research On Image Compression Algorithm Based On All Phase Biorthogonal Transform And All Phase Interpolation

At present, the JPEG still picture compression standard based on DCT has been widely used in image and video compression field. At medium and high bit rates, JPEG compression algorithm shows good compression performance and better reconstructed image. And the complexity of the algorithm is not high, and easy to implement. However, at low bit rates, the reconstructed image has serious blocking artifacts. Based on the idea of all phase, All Phase Biorthogonal Transform (APBT) considers all possible interception phases of the orthogonal transform to reduce the blocking artifacts, and has been successfully applied in image compression. The study of this paper mainly focuses on APBT, all phase filtering, and all phase interpolation applied in image compression.This paper details the theory of APBT and the JPEG-like image compression algorithm based on APBT. In the JPEG-like algorithm, instead of DCT, APBT is used as the new transform. Meanwhile, the uniform quantization interval replaces the complex quantization table. The image compression quality of APBT-JPEG is improved, and at low bit rates blocking artifacts are obviously reduced. However, the simple uniform quantization is not necessarily the optimal quantization, and using nonlinear programming can obtain the optimal quantization matrix to further improve the compression quality. On this basis, APBT is firstly applied in color image compression coding, coding and decoding the separate luminance and chrominance components. The experimental results show that the color image compression algorithm based on APBT outperforms the traditional algorithm.To further improve the compression performance of APBT-JPEG algorithm and reduce blocking artifacts, the original image data is firstly down-sampling at coding end, and then the decompressed data is interpolating to recover the reconstructed image. Before down-sampling all phase DCT low-pass filter is applied, and all phase DCT interpolation is also used as the interpolation method. Finally, simulation results show that better deblocking effect is achieved followed by the comparison and analysis with traditional algorithm.Since the characteristic of the Bayer patterned image, this paper proposes two new algorithms:Bayer patterned image compression based on APBT and structure separation, Bayer patterned image compression based on APBT and structure conversion, in which APBT-JPEG is used instead of the conventional compression algorithm. In the process of interpolation to recover the full color image, all phase IDCT interpolation is adopted followed by the comparison and analysis with conventional interpolation methods. The experimental results show that the algorithms proposed in this paper outperform the conventional algorithm both in subjective and objective aspects.