Radar Image Compression Based On Wavelet Transform And ROI Image Coding

With the development of the maritime transport, the accidents also increased. As the indispensable electronic equipment in shipping industry, radar has become more and more important. Since the radar image has the features of extensive data, high transmission speed, real-time storage requirements and so on, it is important to compress the radar image before transmission. For different characteristics of marine navigation radar and different uses of radar images, sometimes people may only have interest in some specific area around the vessels in the radar image, so it’s important to study the radar image coding based on region of interest. Therefore, the technology of wavelet transform and region of interest about radar image coding is of theoretical and practical significance.According to the wavelet transform and the ROI coding in the JPEG2000standard, the paper put forward the method of SPIHT based on lifting wavelet and ROI image coding techniques.Firstly, the paper discussed the evaluation standard of image coding through the comparison between radar image and natural image, then discussed the character of wavelet’s effect on image compression coding, through the experimental analysis selected suitable wavelet base and decomposition layers, then introduced the embedded wavelet coding, discussed the principle and coding theory of two commonly used coding algorithms, EZW and SPIHT algorithm. Through the analysis of simulation experiment, the paper compared the advantages and disadvantages of both methods and chose the SPIHT algorithm.Finally, the paper mainly introduced the ROI coding algorithm based on SPIHT which improved the coding efficiency through the improvement of the wavelet transform part. The algorithm expanded the wavelet coefficients in the ROI area by way of planar ascension, and the bit rate level of the ROI plane was reduced before the inverse wavelet transform. During this term of coding, no other judgment statement was introduced, and the algorithm complexity is as easy as before.