A Parallelized Approach for Colorizing Grayscale Images

摘要

Image processing plays an integral role in the field of computer science and is incorporated in numerous applications such as remote sensing, medical imaging, image restoration, and machine vision. One of the challenging tasks in image processing is colorizing grayscale images. Restoring color in grayscale images aids in enhancing the visual cognition of image data such medical images, surveillance images, and scientific illustrations. Compared to the relatively simpler process of converting color images to grayscale, the process of colorizing grayscale images is much more complex, is not based on any one particular method, and is significantly dependent on the human perception of color. Additionally, such a colorizing process is generally computationintensive and tedious. Hence, a parallelized highperformance approach based on manycore programming and the CUDA platform is presented in this paper for colorizing grayscale images. The proposed approach has been developed with the objective of incorporation in huge image datasets (such as those obtained from satellites) and video frame sequences.