The proposed authoring method of filtering differs from median filtering and is designed to reduce the impact of impulse noise of different types while maintaining the clarity and quality of the image.
This method is called Least Finite Difference Method (LSD) and consists in replacing the studied pixel with coordinates (x, y) with a new value so that the difference in brightness (or RGB value for color images) its value with possible neighbors was the lowest.
Метою роботи є підвищення ефективності фільтрації зображень при розробці нового методу найменших кінцевих різниць та створення інтерактивного веб-сервісу його реалізації.
The proposed filtering method allows you to effectively eliminate impulse noise from graphic data, taking into account a priori noise information without significant distortion of useful information. The proposed algorithm was used to restore test images, the analysis of digital processing in comparison with other known methods using quantitative and qualitative characteristics.