Naijie Xin, M. Xie, Jiande Shi, Tiantai Lu, Zhifeng Ma
{"title":"Design and Implementation of Target Tracking System in Low Illumination Environment Based on FPGA","authors":"Naijie Xin, M. Xie, Jiande Shi, Tiantai Lu, Zhifeng Ma","doi":"10.1145/3606193.3606199","DOIUrl":null,"url":null,"abstract":"Object tracking has been an important research topic in the field of computer vision. At present, most target tracking algorithms need to work in an environment with good lighting conditions. Environments such as night, rainy days, and foggy days will cause tracking drift and even target loss. In order to solve the above problems, this design proposes a target tracking system that combines image enhancement algorithm and target tracking algorithm. The image enhancement uses the Multi-Scale Retinex (MSR) algorithm to correct the color and dynamic range of the input image; the target tracking algorithm uses the Meanshift algorithm to track the enhanced image. In order to deploy the algorithm to FPGA for edge computing acceleration, a streaming computing architecture is designed, and at the same time, the algorithm is partially refactored at the design level to better adapt to FPGA deployment; finally, high-level synthesis tools are used, combined with optimization instructions A high-efficiency target tracking system with local parallelization and overall pipeline is designed.","PeriodicalId":292243,"journal":{"name":"Proceedings of the 2023 5th International Symposium on Signal Processing Systems","volume":"35 1","pages":"0"},"PeriodicalIF":0.0000,"publicationDate":"2023-03-24","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":"0","resultStr":null,"platform":"Semanticscholar","paperid":null,"PeriodicalName":"Proceedings of the 2023 5th International Symposium on Signal Processing Systems","FirstCategoryId":"1085","ListUrlMain":"https://doi.org/10.1145/3606193.3606199","RegionNum":0,"RegionCategory":null,"ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":null,"EPubDate":"","PubModel":"","JCR":"","JCRName":"","Score":null,"Total":0}
引用次数: 0
Abstract
Object tracking has been an important research topic in the field of computer vision. At present, most target tracking algorithms need to work in an environment with good lighting conditions. Environments such as night, rainy days, and foggy days will cause tracking drift and even target loss. In order to solve the above problems, this design proposes a target tracking system that combines image enhancement algorithm and target tracking algorithm. The image enhancement uses the Multi-Scale Retinex (MSR) algorithm to correct the color and dynamic range of the input image; the target tracking algorithm uses the Meanshift algorithm to track the enhanced image. In order to deploy the algorithm to FPGA for edge computing acceleration, a streaming computing architecture is designed, and at the same time, the algorithm is partially refactored at the design level to better adapt to FPGA deployment; finally, high-level synthesis tools are used, combined with optimization instructions A high-efficiency target tracking system with local parallelization and overall pipeline is designed.