Segmentation of elevation images based on a morphology approach for agricultural clod detection

Olivier Chimi Chiadjeu (1) , Sylvie Le Hegarat-Mascle (2) , Edwige Vannier (1) , Richard Dusséaux (1) , Odile Taconet (1)

1 ESTER - LATMOS
2 IEF - Institut d'électronique fondamentale

This study deals with the segmentation of altitude or elevation images, i.e. images of the distance (푧-coordinate) between the surface or objects and the camera plane. Specifically to our soil science application, these images are acquired on agricultural surfaces in order to evaluate their roughness. The cloddy structure being a key factor to characterize soil roughness, the elevation image analysis aims at detecting and identifying the clods as accurately as possible. Now, rather than defining a new segmentation algorithm, we propose to transform the input data so as to take into account the different criteria characterizing the clod objects, namely the relative altitude and a function of the gradient norm. The proposed approach was validated on three agricultural surfaces (two synthetic and one real) and the results compared to those of an algorithm previously developed specifically for the clod identification problem.

Mots clés: Depth image; Agricultural soil; Cloddiness; Segmentation; Watershed

Domaines: Traitement du signal et de l'image [eess.SP]; Traitement du signal et de l'image [eess.SP]; Sciences de la Terre

Olivier Chimi Chiadjeu, Sylvie Le Hegarat-Mascle, Edwige Vannier, Richard Dusséaux, Odile Taconet. Segmentation of elevation images based on a morphology approach for agricultural clod detection. 5th International Congress on Image and Signal Processing (CISP), 2012, Oct 2012, Chongqing, Sichuan, China. pp.701-705, ⟨10.1109/CISP.2012.6469994⟩. ⟨hal-00797133⟩