A privacy-secure content trading system based on semi-blind fingerprinting is presented. Semi-blind fingerprinting provides privacy-secure content trading as secure as blind fingerprinting at feasible processing cost with sufficient robustness. This system assures a fair trading for both a content provider and a purchaser which is effective for a market where a number of small or not so reliable content providers deal with purchasers. We have been aiming at providing a useful tool for the market by overcoming the following defects. In the basic models of conventional fingerprinting, the user's security could be guaranteed only under the premise that a content provider was perfectly trustworthy. Such premise makes a system unpractical. To overcome this defect, various scheme of blind fingerprinting have been proposed in which cryptography technique is used in order to protect user's privacy. However, these are not practical due to heavy computation cost and insufficient robustness of watermark against manipulations. The semi-blind fingerprinting fulfills the need for both feasibility and robustness by altering encryption with image decomposition that blinds up an image to be unrecognizable. Image decomposition and a customized embedding algorithm are implemented to a web-based system, whose perceptual condition of decomposed images and robustness of watermark is evaluated.