Eliminating Crop Shadows in Video Sequences by Probable Learning Pixel Classification

Autor: Xiaoping Cheng, Tanghai Liu
Rok vydání: 2008
Předmět:
Zdroj: Computer And Computing Technologies In Agriculture, Volume II ISBN: 9780387772523
CCTA
Popis: Shadows have been one of the most serious problems for vegetation segmetation, espescially under conditions of natural random airflow and human or vehicle disturbance. A video sequence processing method has developed in this paper to identify and eliminate crop shadows. The method comprises pixel models and algorithms explained in a probable learning framework. Expectation maximization (EM) for mixture models is established and an incremental EM method is proposed. This method performs a probable reasoning unsupervised classification of pixels for real-time implementation. The results show that the method is quite robust and can successfully remove shadows under natural lighting conditions.
Databáze: OpenAIRE