FPGA implementation of modified fractional wavelet transforms

Abstract

Source coding of robust communication and network transmission of images usually needs the applications of wavelet-based algorithms. Although wireless multimedia sensors are widely used to deliver multimedia content due to the availability of inexpensive CMOS cameras, their computational and memory resources are still typically very limited. It is known that allowing a low-cost camera sensor node with limited RAM size to perform a multi-level wavelet transform, will in turn limit the size of the acquired image. Recently, fractional wavelet filters technique became an interesting solution to reduce communication burden and wireless bandwidth, for resource-constrained devices. Classical fractional wavelet transforms can achieve much reduction in the required memory at the expense of the reconstructed image quality. In this paper, two new modified fractional wavelet transforms are proposed to reduce boundaries artifacts caused by the previous fractional wavelets. In the first proposed technique, the average of the last and the first rows and the average of the row before the last one and the second row of any two successive fractions are obtained and, respectively located at the same rows of the same fractions. In the second proposed technique, the average of the last and the first rows of any two successive fractions is obtained and then, respectively located at the same rows of the same fractions. The resulting reconstructed images of the later technique highlight the promising performance of its applications on different types of images with different sizes using Haar and bio 5/3 wavelet filters. FPGA implementations of such techniques indicate that the later one is also superior.