A crypto-marking method for secure vector map

作者：Jang Bong Joo; Lee Suk Hwan; Lee Eung Joo; Lim Sanghun; Kwon Ki Ryong

来源：Multimedia Tools and Applications, 2017, 76(14): 16011-16044.

DOI：10.1007/s11042-016-3893-1

• Secure vector map
• GIS security
• Perceptual encryption
• Watermarking
• Crypto-marking

A crypto-marking technique can be used to provide a solution to the secure transmission and storage of highly valuable and detailed vector maps. Such a solution would fulfill the requirements for integrity verification, content security, copyright protection, and authentication. However, existing methods cannot ensure the perceptual invisibility of the watermarking in the encryption domain and can leak the encryption key while decrypting a crypto-marked map. In response to these issues, we present a vector map crypto-marking method using both watermarking and progressive perceptual encryption that are independent of each other. The first watermarking step embeds the watermark using the mean of the Euclidian distance (MED) while preserving the object shape and being robust to geometric attacks. The second perceptual encryption step encrypts the progressive regions in a layer, the objects of a region, and the vertices in an object so that a direct random-access object can be enabled in the encryption/decryption processes. To attain the mutual independence of watermarking and encryption, we encrypted the attributes of objects by using random permutations of the position and orientation of the vertices while preserving the SEDs of all the objects. Hence, the watermark can be extracted in a crypto-marked map or progressive encrypted map without decrypting all the objects. From the experimental results, we found that our method is robust to attacks that damage the vector map during the decryption step and does not produce a leakage of the encryption key during the watermark detecting step. Furthermore, we found that the time required by our method is proportional to the number of objects in vector maps, 0.59 [s] for 7,765 objects and 3.52 [s] for 50,679 objects. Our MED-based watermarking takes less time (about 0.07-0.43 [s] than conventional watermarking methods while our progressive encryption takes less time, at about 0.09 [s] - 0.37 [s], than conventional encryption methods.

• 出版日期2017-7