Font Size: a A A

Image Search And Matching System Design And Implementation, On The Dsp

Posted on:2012-03-25Degree:MasterType:Thesis
Country:ChinaCandidate:B YangFull Text:PDF
GTID:2208330332486831Subject:Software engineering
Abstract/Summary:PDF Full Text Request
In traditional industrial areas, the omission of traditional design process or the small changes of environment often cause a gap between the position of the workpiece and the desired results, therefore, a image search and matching system becomes essential which not only has real-time control of the batch process of industrial areas but also has automatic correction of the offset of location. Traditional Harris corner detection method can achieve the purpose of extracting the key points of the workpiece quickly but the matching accuracy is not very high. The SURF algorithm has higher accuracy but the number of extracted feature points is excessive, it can not meet the requirements of industrial real-time. DM6446 is the next generation of embedded devices which aims at the related network video codec applications of TI company designed as a key member of the Davinci family, and it can greatly reduce the embedded video application development cycle and cost.This thesis designed an image search and matching system. In this thesis, the author improves the traditional Harris corner detection algorithm and the SURF feature matching algorithm, the two improved algorithms can enhance the detection accuracy. In order to play their advantages better, the author proposes a combination algorithm to ensure real-time detection and maximize the system's detection accuracy which assembles the improved Harris and improved SURF matching algorithm. This thesis not only achieves the improved Harris and SURF combinatorial algorithm but also achieves the image search and matching system architecture and code migration on the DM6446, and process the code optimization work for the TMS320C64x+'s DSP cores to ensure real-time detection of a system on the DM6446.Experimental results show that the improved Harris and SURF algorithm compared with before improvement both are improved in the detection accuracy, and combinatorial algorithm not only ensure high detection accuracy but also achieve the request of real-time processing pictures. The system's operation shows that the design of combinatorial algorithm porting to DM6446 with processing optimization has the better speed advantage than the un-optimized system.
Keywords/Search Tags:corner detection, SURF, Combinatorial Algorithm, Davinci, DSP
PDF Full Text Request
Related items