Ju Yuan Hsiao

Finding optimal least-significant-bit substitution in image hiding by dynamic programming strategy

Abstract The processing of simple least-significant-bit (LSB) substitution embeds the secret image in the least significant bits of the pixels in the host image. This processing may degrade the host image quality so significantly that grabbers can detect that there is something going on in the image that interests them. To overcome this drawback, an exhaustive least-significant-bit substitution scheme was proposed by Wang et al. but it takes huge computation time. Wang et al. then proposed another method that uses a genetic algorithm to search “approximate” optimal solutions and computation time is no longer so huge. In this paper, we shall use the dynamic programming strategy to get the optimal solution. The experimental results will show that our method consumes less computation time and also gets the optimal solution.

An image copyright protection scheme based on torus automorphism

We propose an approach to protect the copyright of image data. Many watermarking techniques have been proposed to guard the copyright of image data, most of which need to embed the watermark in the cover image. Unfortunately, embedding always results in a reduction of image quality. To achieve the goal of copyright protection without reducing image quality, we incorporate the concept of digital signature into our approach. Firstly, we produce a signature in correspondence with the cover image and the watermark. We send this signature to a certification authority or a trusted third party. If anyone pirates our image, we can still retrieve the watermark image with the signature that was stored at the certification authority and prove that the image is ours. The watermark image that we use is a gray-level image.

A colour image copyright protection scheme based on visual cryptography and discrete cosine transform

Abstract In this paper, a copyright protection scheme based on visual cryptography and discrete cosine transform for colour images is proposed. Note that the watermark emphasized here is a colour watermark. Instead of hiding a watermark in a colour image for identifying the ownership of the image, all that is required is to construct an ownership share according to a master share created from the colour image by applying discrete cosine transform plus a watermark used for copyright certification. By stacking the master share and the ownership share together, the watermark for ownership identification can be revealed through the utilization of visual cryptography plus a little additional computation. Additionally, the present scheme is also extremely suitable for multiple watermarking. The security of this scheme can be guaranteed via a pseudo random mechanism and the feature of visual cryptography. With substantial experimental results, the robustness of this scheme can be demonstrated against various image processing attacks.

A DCT-domain system for hiding fractal compressed images

In this paper, we propose an approach for hiding a secret image in a cover image. In the beginning, the fractal image compression method is used to compress the secret image, and then we encrypt this compressed data by DES. Finally, we embed the encrypted data into the middle-frequency domain of DCT. After embedding the secret image, the goal of steganography can be successfully achieved.

A Color Image Retrieval Method Based on Local Histogram

In this paper, we shall propose a color image retrieval method. Among the methods of color image retrieval, the histogram technique only captures the global properties so that it cannot effectively characterize an image. To overcome this drawback, we propose a scheme to capture local properties so that it can do retrieval more accurately. In our method, we segment the original image into several subimage blocks, and we do color histogram with every subimage block. After combining color histogram vectors into a multi-dimensional vector, we use this multi-dimensional vector to search the database for similar images. The experimental results show that our method gives better performance than the original color histogram technique.

Lossless image compression based on two-way smaller difference

We propose a lossless image compression method, in which we compress an image by saving the differences between the encoding pixels and their adjacent pixels. To deal with a pixel, we refer to not only its up pixel but also its left pixel and replace the value of the pixel itself with the value of difference. The experimental results show that the method gives quite an impressive performance. Since the values of most adjacent pixels in an image are closed, we can use this simple, effective way to compress the image.

A method for protecting digital images from being copied illegally

Due to the recent growth of digital multimedia transmissions on the Internet, the need for the intellectual property rights protection of multimedia information, like images, music and videos, has become more and more important. Many digital watermarking techniques have been proposed to solve this problem by hiding an invisible watermark in an image to prove the ownership of the image. In this paper, however, we propose a new scheme for the copyright protection of images. Our scheme is a DCT-based method and does not actually hide the watermark in the image. In addition, this scheme is extremely applicable for multiple watermarking. According to our substantial experimental results, this scheme is strongly robust against various kinds of image processing attacks.

A hybrid method for estimating the Euclidean distance between two vectors

We propose an alternative method for estimating the Euclidean distance between an image block and a codeword. Our method incorporates the reduced code look-up table (RCLUT) method with an RCLUT-like method to allow alternative selection in the encoding phase in vector quantization (VQ). This method can speed up computation for distance estimation. It provides better image quality than the RCLUT method at the cost of extra storage. According to experimental results, our method provides better image quality than the RCLUT method.

A Novel Scheme for Securing Image Steganography

In this paper, we shall propose a more secure scheme with a larger embedding capacity for image steganography based on vector quantization and discrete cosine transform. Instead of encrypting the whole secret image, our method only encrypts the important features extracted from a secret image and then embeds the encrypted features into a cover image for producing a stego-image. During the work of data encryption and embedment, we introduce the data encryption standard cryptosystem and a pseudo random mechanism to ensure the security of the secret image. Besides, the distortion between the cover image and the stego-image is also imperceptible by the human eye, even when the size of the secret image is as large as that of the cover image. According to the substantial experimental results we shall give later, we can demonstrate the applicability of our method.

A Cross-Type Peak-and-Valley Filter for Error Prevention and Resilience in Image Communications

The peak-and-valley filter has been shown to be a fast and efficient filter in suppressing impulsive noise. In order to find the best filtering route, we have tested several different kinds of scanning orders, such as zigzag scan, Hilbert curve, etc. The result of the tests shows the vertical scanning order is the best one. Next, we have incorporated a simple edge detection scheme into the peak-and-valley filter and then propose a new filter, called the cross-type peak-and-valley filter. Experimental results show that our filter with vertical filtering route outperforms the peak-and-valley filter proposed by Windyga.