Skip to main content
AkademIndex

Products

For developers

AkademBasesoonOpen API for the ecosystem
Latin
English
Article

Object segmentation approach in image processing

Narzullo MamatovTashkent Institute of Irrigation and Agricultural Mechanization Engineers, National Research University, Tashkent, UzbekistanVohid FayziyevTashkent Institute of Irrigation and Agricultural Mechanization Engineers, National Research University, Tashkent, UzbekistanMalika JalelovaTashkent Institute of Irrigation and Agricultural Mechanization Engineers, National Research University, Tashkent, UzbekistanBoymirzo SamijonovSejong University, South Korea, Seoul, Korea
ITM Web of Conferencesjournal2025en
ABI

Abstract

Image segmentation is a crucial and complex process in image processing, fundamental to object recognition. While neural network-based methods are widely used for segmentation, they require substantial resources and are vulnerable to noise and artifacts. This study addresses the need for improved segmentation approaches by proposing a novel four-step sequence with corresponding algorithms for object segmentation in images. The research methodology involves developing a systematic approach to image segmentation, implementing the proposed algorithms, and conducting computational experiments using three distinct image databases. The results of the proposed approaches are compared with those of the DeepLabV3+ Resnet50 model, a deep learning-based image segmentation technique. Our findings demonstrate that the proposed approaches outperform the deep learning model in segmenting untrained objects, while the latter excels only with trained objects. This research contributes to the field by offering a more versatile and robust segmentation method, potentially applicable to a wider range of image processing tasks without the need for extensive training data or computational resources. The study highlights the importance of developing adaptive segmentation techniques that can handle diverse object types efficiently.

Topics

Identifiers

Citations and references

Cited by 031 references
Metrics — AkademScholar · Coming soon