Image Quality Assessment And Improvement Combining Aes- Thetics

With the growing popularity of digital devices and the rapid development of in-ternet, people can easily capture a large number of images. However, a part of images have low quality and can not satisfy people’s needs. On the one hand, effective and objective image quality assessment methods can instruct people to choose high quality images from large amount of images; on the other hand, image quality enhancement can help people optimize existing images. In the research field of image processing, image quality assessment and enhancement also attract more and more attention recent years. Researchers have proposed a series of image quality and enhancement meth-ods. Traditional image quality assessment and enhancement methods mainly concern low level features, like noise, clarity, resolution and so on. But continuous improve-ment of digital devices allow people to get images with high low level features, that is, low-noise, high-definition and high-resolution images. Therefore, image quality assessment and enhancement methods focus on from low level features to high level consideration, such as aesthetics based methods.People mainly consider whether an image meet the subjective aesthetic demands with two aspects. First, whether the subject of the image is clear; second, whether the image composition is friendly to human viewers. In this consideration, we propose an image quality assessment method that combines DoF(depth of field) and composition. We first introduce a blur measurement based low DoF detection approach. A low DoF image with a clear topic and blur background is considered to have great visual impact. Further we assess the image composition by how well it is consistent with the rule of thirds which can make the images more compact. Finally we combine low DoF and composition to get an objective, aesthetic assessment. The experiment results suggest images assessed with high qualities by the proposed method are also good images according to human aesthetics.In addition, we use an content-based image resizing algorithm-Seam Carving to optimize the composition of an image with our composition assessement method. Re-searchers have used cropping-and-retargeting to get better composition. This method is simple and effective, but the improved images lost a lot of information at the same time. Our approach can retain faithful, as much as possible, to the original image content while enhance the image composition. We consider similarity while improve image composition and can get optimizing result image. Experiments show that our algorithm is able to produce the recomposed images with minimal visual distortions with tolerance.