Minimal Memory Consumed Pipeline Filter and Its Implementation on FPGA

Abstract

Pipeline filter is a classical small target detection algorithm for infrared system. However, when the image size is large, the algorithm consumes too many memory resources, which makes it difficult to apply on hardware platforms with limited storage resources. In this paper, a minimal memory consumed pipeline filter is proposed, and its implementation approach based on FPGA is given. This approach stores the measurements list and then restores the original image instead the traditional way of storing the original image directly, which greatly saves the storage resource requirement. A memory consumption comparison between the proposed approach and the traditional approach is given, and the effectiveness of this algorithm is verified on FPGA development board.