Jingbo Wang, Zaifeng Shi, Ke Pang, Tianye Gao, Qingjie Cao
{"title":"A mapping method of image mosaic algorithm on embedded reconfigurable processor","authors":"Jingbo Wang, Zaifeng Shi, Ke Pang, Tianye Gao, Qingjie Cao","doi":"10.1109/CISP.2015.7407995","DOIUrl":null,"url":null,"abstract":"A mapping method of image mosaic algorithm based on Reconfigurable Processing Element Array (RPEA) is proposed. This method can effectively improve the utilization rate of parallel resources. In the image mosaic algorithm, Harris corner detection is used to extract the feature points. Normalized Cross Correlation (NCC) algorithm is used to match these feature points and Random sample consensus (RANSAC) algorithm is used to eliminate the mismatching. Critical parts of the algorithms described in C language are found out, which are dynamically mapped onto the RPEA to run concurrently. The execution time on RPEA is reported to compare with execution time on general single-core processor. This paper focuses on the mapping methods based on loop unrolling and software pipeline loop and proposes an optimized method. The test results show that the speedups of mapping the mosaic algorithm on the reconfigurable processor with the proposed method versus using Intel Atom 230 reach more than 2.","PeriodicalId":167631,"journal":{"name":"2015 8th International Congress on Image and Signal Processing (CISP)","volume":"28 1","pages":"0"},"PeriodicalIF":0.0000,"publicationDate":"2015-10-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":"0","resultStr":null,"platform":"Semanticscholar","paperid":null,"PeriodicalName":"2015 8th International Congress on Image and Signal Processing (CISP)","FirstCategoryId":"1085","ListUrlMain":"https://doi.org/10.1109/CISP.2015.7407995","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
A mapping method of image mosaic algorithm based on Reconfigurable Processing Element Array (RPEA) is proposed. This method can effectively improve the utilization rate of parallel resources. In the image mosaic algorithm, Harris corner detection is used to extract the feature points. Normalized Cross Correlation (NCC) algorithm is used to match these feature points and Random sample consensus (RANSAC) algorithm is used to eliminate the mismatching. Critical parts of the algorithms described in C language are found out, which are dynamically mapped onto the RPEA to run concurrently. The execution time on RPEA is reported to compare with execution time on general single-core processor. This paper focuses on the mapping methods based on loop unrolling and software pipeline loop and proposes an optimized method. The test results show that the speedups of mapping the mosaic algorithm on the reconfigurable processor with the proposed method versus using Intel Atom 230 reach more than 2.
提出了一种基于可重构处理单元阵列(RPEA)的图像拼接算法的映射方法。该方法可有效提高并行资源的利用率。在图像拼接算法中,采用Harris角点检测提取特征点。采用归一化互相关(NCC)算法对特征点进行匹配,采用随机样本一致性(RANSAC)算法消除不匹配。找出了用C语言描述的算法的关键部分,并将其动态映射到RPEA上并发运行。报告RPEA上的执行时间与一般单核处理器上的执行时间进行比较。本文主要研究了基于循环展开和软件流水线循环的映射方法,并提出了一种优化方法。测试结果表明,与使用Intel Atom 230在可重构处理器上映射镶嵌算法相比,该方法的速度提高了2倍以上。