Improved Poisson MAP Algorithm for Better Image Deconvolution

Captured images are usually obtained unclear and blurry. Image deconvolution algorithms are normally applied to get clearer images from their unclear versions. Research related to this field has tremendously grown in the past years due to the increased demand for top-quality images. The maximum a posteriori (MAP) concept has been used previously for image deconvolution. Accordingly, the Poisson MAP is a simple structure iterative algorithm that was proposed for image deconvolution. Iterative means that it needs many repetitions to deliver the output image. The repetitive procedure consumes time and involves more computational costs that can be avoided. Therefore, this algorithm is modified by utilizing two non-complex acceleration factors so that the output is obtained by using fewer iterations, making the algorithm runs faster. The improved algorithm is tested intensively with various unclear images, as well as it is compared with its original version, and the outcomes are evaluated using two evaluation methods (i.e., gradient information with features (GI-F) and run-time). From performing various tryouts, the original algorithm required an average of 36 iterations and an average runtime of 0.49 seconds, while the proposed algorithm required an average of 15 iterations and an average runtime of 0.28 seconds to produce better quality results. Hence, the proposed algorithm has shown better performances than its original version by providing better-acutance results rapidly.