Avaliação da qualidade de Imagens Processadas em Pipelines de Processamento de Sinais por Métodos Cegos

Abstract

The present work aims to analyze the result, in terms of quality, of images processed in pipelines of signal processing of classical architecture images through two blind methods: BRISQUE (Blind/Referenceless Image Spatial Quality Evaluator) and the ∆DoM (Difference of Differences of a Median Filtered Image), both objective methods. Objective methods of quality assessment are characterized by the absence of human observers analysis, returning quality indices. Pipelines for image signal processing are architectures, composed of algorithms, responsible for transforming a raw image signal into an image that meets specific requirements defined by the purpose of the application. In a sample of 8 (eight) different images, the resulting quality of the post-processed images was analyzed and the impact of Linearization and White Balance on the used indices was observed. The impact of these steps was thus observed, and based on that, complementary processes to pipeline were suggested in order to obtain better images acquired by smartphones.