Researches On Scanned Halftone Image Descreening Algorithm

With the rapid development of digital multimedia technology and equipment, people have an increasing higher requirement for the quality of the images. Due to the limits of technology and cost, most of printing devices cannot output real continuous tone of multiple grayscales or colors, but only use the halftone technique to print all kinds of printing products. The digital images obtained by scanning will be corrupted with Halftone Screen Patterns, which is difficult to perceive because of the low resolution of naked eyes. Halftone screen patterns will decrease the vision effect of digital images, and bring a lot of inconvenience for subsequent various applications. Therefore, how to descreen the halftone patterns in scanned images effectively and efficiently, becomes a more important issue of scanned image processing. This thesis will focus on a study of scanned halftone image descreening algorithm.In this thesis, we introduce the research status of inverse halftoning and descreening. Based on the investigation of existing methods on descreening, a fast classification based adaptive descreening algorithm for grayscale scanned halftone image and an image fusion based descreening algorithm for color scanned halftone image are proposed respectively in this thesis. The main contents of this thesis are as follows:1. The imaging process of scanned halftone image is introduced in detail, and the causes of halftone patterns are researched deeply. We analyze the printing and scanning processes step by step to build up an imaging model of scanned halftone image, which is suitable for different printing and scanning devices.2. For grayscale scanned halftone image, in order to recover contone images with nice vision effect from different sources scanned images, and considering the complexity of the algorithm computing, a fast classification based descreening algorithm for grayscale scanned halftone image is proposed in this thesis. The LBP feature and ELM based classifier are adopted to estimate the information of printing and scanning devices, which helps to select adaptive parameters for the following descreening operations. The experiments based on the real scanned images demonstrate that the algorithm can estimate the device information of scanned halftone images well, and the contone images recovered by the following descreening operations are with high image quality. The algorithm is with low complexity and is suitable for real-time processing.3. For color scanned halftone image, a descreening algorithm based on joint filtering is proposed in this thesis. A relative total variation(RTV) operator is used to extract the main structure and edge information of scanned images. And at the same time, a smoothing filter is adopted to smooth the scanned images to obtain a basic recovered image. At last, a joint filter is utilized to fuse the structure image and the basic recovered image. In this step, a rolling mechanism is introduced to control the times of rolling joint filtering to get recovered contone images with different qualities meeting various needs in real life. The algorithm has been evaluated on different color scanned images, and the experimental results demonstrate its effectiveness and robustness.