Registration and Fusion of Close-Range Multimodal Wheat Images in Field Conditions

Multimodal images fusion has the potential to enrich information gathered by multi-sensor plant phenotyping platforms. Fusion of from multiple sources is, however, hampered technical lock image registration. The aim this paper is provide a solution registration and multimodal wheat in field conditions at close range. Eight methods were tested on nadir acquired pair red, green blue (RGB) cameras, thermal camera multispectral array. most accurate method, relying local transformation, aligned with an average error 2 mm but was not reliable for images. More generally, suggested method preprocesses necessary before (plant mask erosion, pixel intensity averaging) would depend application. As consequence, main output study identify four registration-fusion strategies: (i) REAL-TIME strategy solely based cameras’ positions, (ii) FAST suitable all types tested, (iii) (iv) ACCURATE HIGHLY strategies handling distortion unable deal very different natures. These suggestions are, limited compared study. Further research should investigate how recent cutting-edge perform specific case canopy.