Search

The human visual system unconsciously determines the color of the objects by "discounting" the effects of the illumination, whereas machine vision systems have difficulty performing this task. Color constancy algorithms assist computer vision pipelines by removing the effects of the illuminant, which in the end enables these pipelines to perform better on high-level vision tasks based on the color features of the scene. Due to its benefits, numerous color constancy algorithms have been developed, and existing techniques have been improved. Combining different strategies and investigating new methods might help us design simple yet effective algorithms. Thereupon, we present a color constancy algorithm based on the outcomes of our previous works. Our algorithm is built upon the biological findings that the human visual system might be discounting the illuminant based on the highest luminance patches and space-average color. We find the illuminant estimate based on the idea that if the world is gray on average, the deviation of the brightest pixels from the achromatic value should be caused by the illuminant. Our approach utilizes multi-scale operations by only considering the salient pixels. It relies on varying surface orientations by adopting a block-based approach. We show that our strategy outperforms learning-free algorithms and provides competitive results compared to the learning-based methods. Moreover, we demonstrate that using parts of our strategy can significantly improve the performance of several learning-free methods. We also briefly present an approach to transform our global color constancy method into a multi-illuminant color constancy approach. [ABSTRACT FROM AUTHOR]

Mapping hazelnut orchards can facilitate land planning and utilization policies, supporting the development of cooperative precision farming systems. The present work faces the detection of hazelnut crops using optical and radar remote sensing data. A comparative study of Machine Learning techniques is presented. The system proposed utilizes multi-temporal data from the Sentinel-1 and Sentinel-2 datasets extracted over several years and processed with cloud tools. We provide a dataset of 62,982 labeled samples, with 16,561 samples belonging to the ‘hazelnut’ class and 46,421 samples belonging to the ‘other’ class, collected in 8 heterogeneous geographical areas of the Viterbo province. Two different comparative tests are conducted: firstly, we use a Nested 5-Fold Cross-Validation methodology to train, optimize, and compare different Machine Learning algorithms on a single area. In a second experiment, the algorithms were trained on one area and tested on the remaining seven geographical areas. The developed study demonstrates how AI analysis applied to Sentinel-1 and Sentinel-2 data is a valid technology for hazelnut mapping. From the results, it emerges that Random Forest is the classifier with the highest generalizability, achieving the best performance in the second test with an accuracy of 96% and an F1 score of 91% for the ‘hazelnut’ class. [ABSTRACT FROM AUTHOR]