A novel subthreshold-region approximate edge detection structure composed of two new inexact full adders applicable for sensitive bioimages

Abstract Edge detection in medical images plays an important role in diagnosing diseases. This research presents a novel approximate edge detection architecture based on new approximate full adders (FAs). Two subthreshold approximate FAs (AFAs) are proposed, named proposed AFA (PAFA1) and PAFA2, and implemented by 90 nm complementary metal-oxide semiconductor technology. A voltage supply ( V DD ) of 0.3 V satisfies the subthreshold region and reduces the power consumption of the edge detection unit. In the PAFA1, output carry ( C out ) is directly derived from input B with an error rate (ER) of 75%, while in the PAFA2, C out is sourced directly from input carry ( C in ) with an ER of 87.5%. The total areas of PAFA1 and PAFA2 are 44.34 µ m 2 and 112.18 µ m 2 , respectively. Incorporating PAFA1 makes a low-power (0.28 µ W) and high-speed (delay of 0.28 ns) edge detection structure. This configuration significantly reduces the power-delay product and energy-delay product. Additionally, PAFA1 features the lowest transistor count, with only 32 transistors, which further minimizes the power-delay-area product and energy-delay-area product. Conversely, PAFA2 in the edge detection architecture delivers optimal results of accuracy and peak signal-to-noise ratio, 0.9532 and 13.2961, respectively. The proposed edge detection system containing new AFAs is reliable in bioimage processing.