Real-time image detail enhancement implementing on multi-core DSP platform

Abstract

Image detail enhancement is of wide application in image processing fields. In order to gain a better performance for detail enhancement, the computing cost is normally very large. Thus, it's unacceptable in real-time applications, such as video enhancement. In this paper, we propose a novel parallel framework for reliable real-time image detail enhancement based on recent proposed Gradient Domain Bilateral Filter (GDBF) which is realized on a multicore DSP platform: TMS320C6678. With the implement of parallel enhancement method, not only image base layer and detail layer are rapidly separated, but also the detail layer is further enhanced. Furthermore, the implementation of the parallel framework employs fix point programming instead of float point. With a bit-shift strategy for input image data, the computing precision is well satisfied. Experimental results show that our proposed framework outperform available state-of-the-art one, the processing time for 1080p image is less than 30ms. It is a good candidate for real-time image detail enhancement based on the multi-core DSP.