In the traditional vector quantization (VQ) encoding procedure, the quality of all parts of the image is evenly distributed. The code words selected for encoding depend on the complexity of the image and quality requirements. However, VQ cannot be used to enhance the quality of only the regions of interest (ROI). Therefore, an improved weighted cell-split algorithm based on weighted ROI blocks is proposed to train a code book to generate improved code words with emphasis on the ROI blocks. Furthermore, code words can be manipulated to allow the user to define the percentage of ROI emphasis. Experimental results show that the image quality of ROIs is significantly improved by more than 1 dB peak signal-to-noise ratio. Moreover, the improved cell-split algorithm can be easily integrated into the traditional VQ coding procedure since only the contents of the code book are modified. The algorithm can be applied to medical images where the integrity of the ROI blocks is important.
We introduce a new secret sharing scheme, where the novelty lies in the use of a fixed angle segmentation technique to create circular shadow images called shares. In this way, the shares can be stacked in different angles to reveal different secret messages. The participants must be in possession of both the shares and the stacking angles, which creates the additional degree of protection.
Due to the dramatic growth of network and multimedia technology, people can more easily get variant information by using Internet. Unfortunately, it also makes the diffusion of illegal and harmful content much easier. So, it becomes an important topic for the Internet society to protect and safegurard Internet users from these content that may be encountered while surfing on the Net, especially children. Among these content, porno graphs cause more serious harm. Therefore, in this study, we propose an automatic system to detect still colour porno graphs. Starting from this result, we plan to develop an automatic system to search porno graphs or to filter porno graphs. Almost all the porno graphs possess one common characteristic that is the ratio of the size of skin region and-skin region is high. Based on this characteristic, our system first converts the colour space from RGB colour space to HSV colour space so as to segment all the possible skin-colour region from scene background. We also apply the texture analysis on the selected skin-colour region to separate the skin region from non-skin region. Then we try to group the adjacent pixels located in skin region. If the ratio is over a given threshold, we can tell if the given image is a possible porno graph. Based on our experiment, less than 10% of non-porno graphs are classified as pornography, and over 80% of the most harmful porno graphs ara classified correctly.
In this paper,we propose an automatic image processing system to beautify human faces in frontal-parallel color images. Although most of image processing packages provide functions to beautify color images, a few of them, at least based on our knowledge, are specific for beautifying human faces. By using these functions, the processed face images become unreal. For example, they will remove most of natural edges around some special regions, such as eyes and mouth. Therefore, the proposed system only processed the regions of faces. To make the processed face region smoother, our system treats the regions of eyes/mouth and the rest of face region differently. By using different methods to smooth these two types of regions, we can keep almost all the natural edges around eyes and mouth , but remove wrinkles and spots on the rest of faces. The process of our system ia as follows. At beginning, we convert the RGB color space into YCbCr space so as to segment face region from scene background based on the value range of the skin color proposed by H.A. Rowly, etc. Within the face region, the system uses the chain-code to get the eye region and the mouth region. For the eye and mouth regions, we adjust the image pixels by pixels; the rest of pixels are justified by block base. To evaluate the performance of our system, we compare our system with the tool?cleanSkinFX, which can be found at the Web site http://www.mediachance.com. Our system is outperforming.
This paper describes a fast and effective approach for fingerprint image preprocessing. It is suitable for the minutiae matching because the approach could filter out error skeletons in the fingerprint image. Traditional preprocessing of fingerprint recognition uses quantization interval to adjust the pixel values in a gray-level fingerprint image to clear the fingerprint ridges. Then, it applies eight direction windows to modify the fault direction of the fingerprint ridges. The third stage is converting the gray-level fingerprint image to black-while one, and thins the fingerprint ridges to their skeletons with one-pixel depth. However, the above procedures may also result error skeletons, such as branches, noises, and gaps. We therefore need to filter those errors out to raise the recognition rate. Experimental results show that our approach is not only simple and fast, but also has the ability to delete all kinds of error skeletons. Hence the approach suggested in the paper should be appropriate for the minutiae matching.