红外技术
紅外技術
홍외기술
INFRARED TECHNOLOGY
2013年
10期
632-637
,共6页
杨陈晨%顾国华%钱惟贤%陈钱%徐富元
楊陳晨%顧國華%錢惟賢%陳錢%徐富元
양진신%고국화%전유현%진전%서부원
红外图像配准%角点检测%KLT跟踪%FPGA
紅外圖像配準%角點檢測%KLT跟蹤%FPGA
홍외도상배준%각점검측%KLT근종%FPGA
infrared image registration%corner detection%KLT tracking%FPGA
提出了一种基于Harris角点的KLT跟踪图像配准算法用于解决红外搜索跟踪系统中红外图像配准问题。该算法首先对红外序列图像进行Harris角点检测,选取合适的角点作为跟踪点,对跟踪点进行KLT跟踪,通过KLT算法计算出序列图像间的位移量,从而实现红外图像配准。整个算法在DE2-115系统平台上实现,采用CycloneIV系列的FPGA,在Quartus13.0软件上利用Vhdl编写和调试Harris角点检测、KLT跟踪以及匹配程序。最后结论表明,该算法能满足红外图像配准的精度和稳定性,并且FPGA在图像处理中可以提供更高的运算速度,更能满足图像配准的实时性要求。
提齣瞭一種基于Harris角點的KLT跟蹤圖像配準算法用于解決紅外搜索跟蹤繫統中紅外圖像配準問題。該算法首先對紅外序列圖像進行Harris角點檢測,選取閤適的角點作為跟蹤點,對跟蹤點進行KLT跟蹤,通過KLT算法計算齣序列圖像間的位移量,從而實現紅外圖像配準。整箇算法在DE2-115繫統平檯上實現,採用CycloneIV繫列的FPGA,在Quartus13.0軟件上利用Vhdl編寫和調試Harris角點檢測、KLT跟蹤以及匹配程序。最後結論錶明,該算法能滿足紅外圖像配準的精度和穩定性,併且FPGA在圖像處理中可以提供更高的運算速度,更能滿足圖像配準的實時性要求。
제출료일충기우Harris각점적KLT근종도상배준산법용우해결홍외수색근종계통중홍외도상배준문제。해산법수선대홍외서렬도상진행Harris각점검측,선취합괄적각점작위근종점,대근종점진행KLT근종,통과KLT산법계산출서렬도상간적위이량,종이실현홍외도상배준。정개산법재DE2-115계통평태상실현,채용CycloneIV계렬적FPGA,재Quartus13.0연건상이용Vhdl편사화조시Harris각점검측、KLT근종이급필배정서。최후결론표명,해산법능만족홍외도상배준적정도화은정성,병차FPGA재도상처리중가이제공경고적운산속도,경능만족도상배준적실시성요구。
In order to solve the infrared image registration problem of the infrared seeking and tracking system, an infrared image registration algorithm which is based on the Harris corner of KLT tracking is put forward. The algorithm firstly does Harris corner detection of infrared image sequences and choose suitable corners as tracking points. Then the tracking points are tracked by KLT and the displacement between image sequences are calculated by the KLT algorithm. Thus infrared image registration is realized. The whole algorithm is implemented on the DE2-115 system platformwith CycloneIV series FPGA of which programming and debugging in Harris corner detection, KLT tracking and matching process is completed by Vhdl language in Quartus 13.0 software. Final conclusion shows that the proposed algorithm can satisfy the precision and stability of the infrared image registration, and the FPGA in the image processing system can provide higher speed, meeting the real-time requirements of image registration much more.