Evaluation of the Effectiveness of Interpolation Methods in the Process of Image Size Standardization

In order to speed up the process of analyzing images and increase the accuracy of the analysis result, at the stage of pre-processing images, image resizing plays a key role. Various interpolation methods are used to calculate new pixel values when changing image size. In this research work, various methods of interpolation, both adaptive and non-adaptive, were studied. Image interpolation is used in image sampling to reduce the difference in quality to the original image. During the study, the nearest neighbor, Billinear, bicubic, Lanczos, Area, NEDI and similar methods of interpolation were studied and their effectiveness was evaluated and comparative analysis was carried out. Metrics such as MSE, PSNR, and SSIM were used to evaluate interpolation results. According to the results obtained, when interpolating images, the bicubic interpolation method showed the highest results (MSE=5.99, PSNR=40.35 dB, SSIM=0.96). This research work is expected to serve as the basis for the initial processing phase of images before transferring them to machine learning and in-depth training algorithms for image analysis.

Ҳали таржима қилинмаган