| In many modern applications of multimedia information systems, such as video teleconferencing, medical imaging, and military applications, the high-resolution image frames contain large amounts of image data. These images require large amounts of memory for storage, and high-speed networks for transmission. Hence, it becomes necessary to compress these images to reduce the required storage space and to increase the network transmission speed. This is particularly critical since image size continues to increase along with the desire for better image quality, resolution, and integrity. Studies in alternative means for bandwidth conservation are driving the renewed efforts for improvements in video and image compression. Prominent among the improved methods of image compression is the wavelet transform.; Wavelet transforms present certain inherent advantages when employed in image compression, especially when implemented in hardware. The simplest transform, the Haar Transform, yields fair compression ratios and reconstructed image quality, while providing extremely high performance due to its computational simplicity. When applied to a two-dimensional (2-D) signal such as an image, the transform typically operates first along every row, and then along every column of the partially transformed image. It is possible to algebraically reduce this process into a single transform, thereby saving much time and memory and further exploiting the simplicity of the Haar Transform. This improved transform is hereafter dubbed the Super-Efficient Haar Transform, or SEHT.; This body of work describes the design, testing, and final implementation of an entire wavelet-based color image compression system utilizing the SEHT. It describes three stages of development: high-level algorithmic development in Matlab, mid-level hardware simulation in Matlab, and a low-level VHDL implementation developed from the hardware simulation. The final product is a functional image compression system implemented on Field Programmable Gate Arrays (FPGAs) that will fully compress and decompress color, still images. |
