Research On Digital Watermarking For Remote Sensing Image Based On Feature Region

Remote sensing image is one of the main data sources of geospatial information t has the characteristics of obvious time, rich data and fast update rate, it provid unprecedented mass information for geoscience research, and also provides import data support for the country’s economic development and national defen construction. With the integration of network technology, open platform for informatio publishing, and high degree of data sharing, remote sensing image data inevitably faces with unsafe problem.As a new technology to solve the problem of copyright protection of digital product digital watermarking technology is an effective technical means for multimedia data file authentication, anti-counterfeiting, tamper-proof, data security and integrity protection Although remote sensing image data with high accuracy, as far as the current technoloev level, there is still some redundancy, and the limitation of human perception also provides the possibility for the watermark hidden.Therefore, digital watermarking for remote sensing image provides a viable solution for copyright attribution.According to the remote sensing image data’s unique attributes and characteristics data processing, interpretation accuracy requirements and application requirements the main research content in this paper is as follows:(1) It analysis of the whole process of remote sensing data, determines the timing relationship of digital watermarking embedding and image processing, clears the major attacks that watermark image faces, so that to select the digital watermarking embeddina feature region in remote sensing image and preprocess the watermark for attacks(2) It proposes watermarking means for remote sensing image based on Harris-Laplace featuue region. It generates a pseudo-random sequence bv pseudo-random sequence generator, and selects one paragraph of the values as watermark information after binarization; then it embeds the watermark into spatial domain after dividing the extracted Harris-Laplace feature region into rings in accordance with the watermark.It scrambles the Binary watermark image, then scans it into one-dimensional vector as watermark information; it normalizes the extracts Harris-Laplace feature region in the host remote sensing image, and applies Contourlet transformation with L Scale and λ direction, then it embeds watermark in the selected sub-band bandpass coefficients by odd-even quantization method.(3) It proposes watermarking means for remote sensing image based on SIFT featuue region.It divides the extracted SIFT feature region into sectors in the host remote sensing image, then embeds the pseudo-random sequence watermark into spatial domain.It normalizes the extracts SIFT feature region in the host remote sensing image, applies Contourlet transformation with L Scale and λ direction, and then it embeds the scrambled Binary image watermark into the selected sub-band bandpass coefficients by odd-even quantization method.(4) It does experiments of digital watermarking embedding and extraction, watermark transparency, resistance to conventional signal processing and geometric attack robustness, the statistical characteristics and application effect comparison of the original image and the watermark images with the watermarking algorithms for remote sensing image based on Harris-Laplace feature region in spatial domain and Contourlet domain, the watermarking algorithms for remote sensing image based on SIFT feature region in spatial domain and Contourlet domain,in order to verify the correctness and validity of the methods in this paper.