Graph-based image gradients aggregated with random forests

• Published in: Pattern recognition letters Vol. 166; pp. 182 - 189
• Main Authors: Almeida, Raquel, Kijak, Ewa, Malinowski, Simon, Patrocínio Jr, Zenilton K.G., Araújo, Arnaldo A., Guimarães, Silvio J.F.
• Format: Journal Article
• Language: English
• Published: Elsevier B.V 01.02.2023; Elsevier
• Subjects: Computer Science; Edge detection; Graph; Hierarchical watershed; Random forest; Segmentation; Random forest; Graph; Segmentation; Hierarchical watershed; Edge detection; segmentation; edge detection; Hierarchical Watershed; graph; Random Forest
• ISSN: 0167-8655; 1872-7344
• DOI: 10.1016/j.patrec.2022.08.015

•Grouping pixels reduced time and performance on edge detection and segmentation.•Expanding the analysis region yielded similar results with the original proposal.•The position of the features is important for the edge detection with random forest.•Sharp thick contours, uniform regions and small details impact the final segmentation.•Statistical analysis demonstrated superiority in segmentation over most edge maps. Gradient methods subject images to a series of operations to enhance some characteristics and facilitate image analysis, usually the contours of large objects. We argue that a gradient must show other characteristics, such as minor components and large uniform regions, particularly for the image segmentation task where subjective concepts such as region coherence and similarity are hard to interpret from the pixel information. This work extends the formalism of a previously proposed graph-based image gradient method that uses edge-weighted graphs aggregated with Random Forest (RF) to create descriptive gradients. We aim to explore more extensive input image areas and make changes driven by the RF mechanics. We evaluated the proposals on the edge and segmentation tasks, analyzing the gradient characteristics that most impacted the final segmentation. The experiments indicated that sharp thick contours are crucial, whereas fuzzy maps yielded the worst results even when created from deep methods with more precise edge maps. Also, we analyzed how uniform regions and small details impacted the final segmentation. Statistical analysis on the segmentation task demonstrated that the gradients created by the proposed are significantly better than most of the best edge maps methods and validated our original choices of attributes.