Bit-depth expansion and tone mapping

Abstract

Bit-depth conversion is very important in the field of image processing. This thesis proposes a method of bit-depth up conversion and a method of bit-depth down conversion. In many applications, it is necessary to increase the bit-depth of images by expanding p-bit integer values to q-bit integer values (q > p). For example, if the bit-depth of an image is lower than that of the display, some contour-like edges will be appeared in the region where pixel values are smoothly varying each other. Thus, bit-depth expansion is important for displaying a low bit-depth image in a high bit-depth monitor. Existing methods tend to give disturbing contouring or blurring artifacts. As the luminance ratio is very large in reality, high dynamic range (HDR) images are suggested to be generated for capturing images. However, because of the limited contrast ratio of common displays and printers, tone mapping (a kind of bit-depth down conversion) is required to convert HDR images back to low dynamic range (LDR) images for displaying purpose. Existing tone mapping methods usually provide good results, but original image's characteristic has been violated at most of the time. In this thesis, for bit-depth up conversion, a novel, simple and efficient adaptive method is proposed to increase the bit-depth in order to take advantage of the existing techniques to give superior image quality. For bit-depth down conversion, a novel and efficient tone mapping algorithm is proposed to convert HDR images to LDR images. Adaptive techniques are also manipulated to provide better visual quality. The whole process is automatic and no parameter is required for manual input. The result will be a superior visual quality tone-mapped LDR image with original HDR image's characteristics.