A Novel Approach for Generating Composite Sketches from Mugshot Photographs

• Published in: 2018 International Conference on Advances in Computing, Communications and Informatics (ICACCI) pp. 460 - 465
• Main Authors: Pallavi, S., Sannidhan, M.S., Sudeepa, K.B., Bhandary, Abhir
• Format: Conference Proceeding
• Language: English
• Published: IEEE 01.09.2018
• Subjects: Enhancement of the sketch; Face; Face recognition; Image color analysis; Image quality; Image recognition; Lighting; Measurement; MSE (Mean Squared Error) and NAE (Normalised Absolute Error); PSNR (Peak Signal to Noise Ration); Quality metrics; Sketch generation
• DOI: 10.1109/ICACCI.2018.8554564

Generation of sketch is being considered as a prominent research topic in image processing, the importance of which is increasing manifold. Converting an image from pixels to a sketch has a massive effect on several applications. Sometimes, the generated sketches may appear blurred and of degraded quality, then, the enhancement of sketch becomes a promising approach to increase each detail of the sketch. Measuring the quality of the generated sketch is important as the performance solely depends on image quality. In this paper, we develop a novel approach to generate the sketch for a corresponding input facial image by undergoing several stages. Here, Gaussian blur is used to blur the given image. Subsequently, enhancement of the sketch is done using Unsharp masking technique. We measure the quality of sketch before and after enhancement using quality assessment methods such as PSNR, MSE, NAE and it is compared against benchmarked quality measures. Experimental analysis is conducted using CUHK and PRIP-HDC datasets.