Research On Image Acquisition Based On STM32 And Its Registration Technique

With the rapid development of the ARM technology,the technology of image acquisition and its processing has deeply influenced our daily live,which is widely used in various fields including production,live and scientific research etc.It makes image acquisition and processing system possible to be miniaturization,low cost and low-power dissipation by the emerging of STM32 chips.The image acquisition system based on STM32 and its registration aims to acquire image data by STM32 combined with corresponding functional modules and then get the images registered.The technology of image registration is the foundation to some practical technology like image fusion,image mosaic,objective detection and identification etc.On this occasion,a design of image acquisition based STM32 and its registration is proposed in this paper.The whole system can achieve the goals of image acquisition,display,storage and image registration.For the image acquisition system,in view of the demand of the system and the performance feature of STM32,the overall design is divided into two parts,one of which is hardware design,and the other is program design.The hardware design includes main-control module design,camera module design,TFTLCD display module design and SD card storage module design.The system chooses STM32F407ZGT6 based on Cortex-M4 core as the main-control device.The program design of the system was designed into modularization,mainly including camera module program,SD card storage module program and TFTLCD display module program.The objective of the image registration part is making the acquired images registered.The image registration is based on SURF algorithm.First of all,the paper detected and located interest points by SURF algorithm.Secondly,it found out the corresponding point pairs in the unregistered image and the reference image,and then worked out the mapping relation between the two images by RANSAC algorithm and least square method,filtered out the mismatching point pairs.Finally,made interpolation to images with the transform parameter acquired above.The paper verified the effectiveness of the whole algorithm on MATLAB at last,providing foundation for image fusion and image mosaic etc.The image acquisition system designed by this paper has advantages like miniaturization,low cost,low-power dissipation and stable,and achieves the goals which set at first.SURF performance well on accuracy and computing time,which has a widely using prospect in fields like image fusion,image mosaic,objective detection and identification and so on.