粮虫在线检测图像的Daubechies小波压缩算法  

AN IMPROVED DAUBECHIES IMAGE WAVELET ALGORITHM RESEARCH IN ONLINE GRAIN PEST INSECT DETECTION

作　　者：廉飞宇 付麦霞[2,3] 许德刚 

机构地区：[1]粮食信息处理与控制教育部重点实验室,450001 [2]粮食光电探测与控制河南省重点实验室,450001 [3]河南工业大学信息科学与工程学院,450001

出　　处：《粮食储藏》2018年第1期17-22,共6页Grain Storage

基　　金："十三五"国家重点研发计划项目(2017YFD0401003)

摘　　要：当采用嵌入式设备实现一个在线粮虫图像检测系统时,由于嵌入式系统硬件资源的不足及存储空间的限制,传统图像压缩算法通常无法满足实时图像处理的要求。针对这些问题,在ARM9 S3C2440嵌入式平台上提出了一种改进后的新型小波变换算法。通过改变传统算法过程中浮点数据的整型化,在有效保留传统小波算法优点的同时降低了算法复杂度。本文改进小波三级压缩算法与传统JPEG算法的在S3C2440嵌入式平台上的比较试验表明,在压缩时间、峰值信噪比等主要性能防线,本文算法均优于传统算法,非常适用于采用嵌入式平台的实时粮虫图像处理系统。Due to the limit of hardware resources and memory space, in the embedded image processing filed, tra- ditional image compression algorithm usually cannot meet the demands of processing grain pest insects image at real - time. In view of these problems, a new improved wavelet transform algorithm is proposed on the ARM9 embedded platform. Through changing the floating -point data from traditional method into integer, this method effectively reserves the advantages of traditional wavelet algorithm, and also reduces algorithm complexity. Through the design of comparative experiments with the traditional JPEG algorithm, we compared the parameters of these two algorithms including compression time, ratio and PSNR. Experimental re- sults based on S3C2440 platform show, compared to traditional JPEG algorithm, on main parameters such as compression time and PSNR, the algorithm proposed in this paper has better performance, which make it to be more adaptable to grain pest insects image processing on ARM platform.

关 键 词：嵌入式 图像压缩 DAUBECHIES小波 改进小波算法 峰值信噪比 

分 类 号：TP391.41[自动化与计算机技术—计算机应用技术]