Research On Wavelet Transformation And Edge Detection Algorithm Based On Embedded Image Acquisition System

With the rapid development of Information Communications and Internet technology, digital information processing capacity becomes larger and larger. However, digital images, which make up 80 percent of data information, become the main part of the information processing. Especially, in the mobile image surveillance and acquiring filed which based on embedded system, since the limitations of hardware resources and mobile networks, how to reduce the amount of data of storage and transmission in order to reduce the amount of information became the key problem which needs urgent solutions on this platform.Image wavelet analysis theory was developed in the past two decades from the Fourier transformation, which is a new digital image processing method. It has a good time and frequency domain analysis features, but since the limitations of the hardware resources on embedded platform, the traditional wavelet transformation which has a high algorithm complexity can not be well carried out on this platform. In order to solve these problems, this paper innovate the traditional wavelet algorithms. A new improved lifting wavelet transformation algorithm based on ARM9 embedded platform is proposed. Additionally, before and after wavelet processing, the edge information of the image was detected. This paper mainly includes the following aspects:1. Research the characteristics of an embedded image acquisition system, set up an experimental platform based on AMR9. And rationally configure the platform hardware and software environment, comply the operating system kernel and change the stream interface driver configuration of the CMOS sensor.2. Research the image wavelet transform coding theory, discuss the wavelet compression and wavelet reconstruction in the two-dimensional image signal processing application, and discuss the features and contrasts of the wavelet function. And then a simulation experiment of the traditional wavelet image compression and reconstruction is carried out based on Daubechies Wavelet function which used in this paper. 3.Analyse the transplantation of the wavelet transform algorithm on the embedded platform , and choose a reasonable wavelet function. An innovative algorithm based on lifting integer wavelet algorithm is proposed, which based on db9/7 wavelet functions. The algorithm not only effectively retains the excellent features of traditional wavelet transform algorithm such as good time-frequency characteristics, multi-resolution analysis, but also reduces the complexity of the algorithm. At last, compared with the traditional JPEG algorithm, an comparative experiment is designed. Experimental results show, under the same compression ratio, the 3-level wavelet algorithm shortens processing time about 40%, and also improves the PSNR about 10%. It is especially suitable for embedded real-time image processing systems.4. In order to reflect the outline and noise information changes before and after the wavelet image processing, a comparative experiment of edge detection is carried out based on classic algorithm. The results show that improved wavelet image compression algorithm has a good de-noising and excellent feature to restore the original image information, and also the paper provides the most suitable edge detection algorithm for this system.