Obfuscated Hardware Accelerators for Image Processing Filters - Application Specific and Functionally Reconfigurable Processors

Abstract

Hardware accelerators are widely used as computationally-intensive cores in consumer electronics (CE) applications. However security and speed of such hardware accelerators, that are responsible for computing data-intensive tasks, play an important role in improving consumer experience in terms of safety and performance. This article presents novel low power multi-modal hardware accelerator architectures viz. application specific processor and functionally reconfigurable processor for image processing filter of $3\times 3$ kernel matrix size. In the proposed functionally reconfigurable processor of $3\times 3$ filter, the same design can be used for five different image processing filters - blurring, sharpening, vertical embossment, horizontal embossment and Laplace edge detection, by varying control input. Further, application specific processor designs of these five types of $3\times 3$ filters are also presented in this article. Additionally, application specific processor architecture of $5\times 5$ filter kernel matrix size is also reported in this article. The results confirm that the proposed hardware accelerators achieve strong security and low design cost. © 1975-2011 IEEE.