Thresholding of Benchmark Images
Venkatesan Rajinikanth, Nadaradjane Sri Madhava Raja, Nilanjan Dey · 2020
The computer algorithm based thresholding is widely discussed in the literature due to its merit and speed of response. In this approach, the threshold identification procedure is implemented using a chosen HA with a chosen OF .After developing an appropriate threshold selection technique, its performance is to be initially assessed using some well-known benchmark images. The computer algorithm developed for a chosen image modality (grayscale/RGB) is initially used to process benchmark images of a chosen size. The results attained with benchmark images are assessed based on their image quality values (IQV). If the attained IQV for bilevel and multilevel operation is superior, then the proposed scheme can be considered to examine real test images, which need to be enhanced. If the IQV attained with the benchmark images are poor, the proposed algorithm needs fine-tuning and this may be achieved by replacing the HA and OF with a new one as per the suggestions provided in the literature. In this chapter, the benchmark grayscale images of dimensions 481 × 321 × 1, 321 × 481 × 1, and 512 × 512 × 1 are considered for discussion, and a MATLAB software simulation is executed using the bilevel and multilevel thresholding with a single OF (Otsu/Kapur) and a chosen HA (CS).