Power-Optimized Gaussian Filtering with Approximate Adders and logical optimization

Gaussian filters are crucial in numerous image and signal processing applications for tasks like noise reduction and smoothing. In resource-constrained environments, such as embedded and mobile devices, optimizing these filters for power efficiency is paramount. This paper focuses on designing a power-efficient Gaussian filter architecture through Design Space Exploration (DSE). The approach leverages logical optimizations and introduces approximate adders to minimize power consumption without significantly impacting performance. The trade-offs between power, area, and output quality are analyzed, with results demonstrating significant power savings while maintaining an acceptable level of image quality.