In this paper, we presents a framework for hardware accelerating methods for video post processing system implemented on FPGA. The histogram equalization and image-sharpening filter are the algorithms ported onto the SOC having Altera NIOS-II soft processor. Custom instructions are chosen by identifying the most frequently used tasks in the algorithm and the instruction set of NIOS-II processor has been extended. Saturation and the Histogram are the new instructions added to the processor that are implemented in hardware and interfaced to the NIOS-II processor. The software implementation of the algorithms has been modified to use the new instructions added for computing the saturation and histogram calculation. Performance of the software implementation of the histogram equalization and sharpening algorithms is boosted by these new instructions added to the NIOS-II processor. The comparison shows that the implemented tasks have been accelerated by -multiple‖ times. The saturation instruction is generic instruction, which can be used in many Image processing applications. The benefit of speed is obtained at the cost of very small hardware resources.

Robust watermarking proposals supported on human visual characteristics with a series of hybrid transform of type discrete wavelet transform (DWT) followed by singular value decomposition (SVD) is wished-for. By analyzing the matrices U or V through SVD, it is bringing into being that there stay alive a well-built relationship amid the internal column elements of U or internal row elements of V. Hence, this work will make the most of these chattels for image watermarking. At the outset, visual digital data is segregated into 8 × 8 non-overlapping pixel blocks and each block is processed for brinks by using the algorithm of detection for a canny brink. An appropriate block is decided to pick in such a way that the number of brinks in each block is only about or equal to a threshold. A threshold is defined by finding the mean of the brinks in each block of the host visual digital data. Using these appropriate blocks, we will form an image of reference. This reference image is processed by a series of operations DWT-SVD. Then, the watermark is implanted by adapting the nth column of the U matrix of the host image with the nth column of the U matrix of the watermark image. The same operation is applied on the V matrix instead of a column vector, use a row vector. The adapted relation is wont to retrieve a watermark. The experimental findings demonstrate that the ideal watermarking algorithm will guarantee that the typical image processing operations and geometric attacks are invisible and more stable. The efficiency of this proposed method is out of shape than other proposed methods examined in this research.