Automated canopy temperature estimation via infrared thermography: A first step towards automated plant water stress monitoring

Canopy temperature estimation is an important process for plant water status monitoring. In this paper the problem for measurement data of the scene acquired via an infrared (IR) thermography system is considered. An optical image taken from the plant canopy is aligned with the underlying IR image, so that plant leaf area can be extracted via simple colour identification techniques and then the temperature distribution of the leaf area is obtained. The success of this procedure relies on the assumption that both optical-IR image alignment and leaf area extraction are perfect. In practice, such assumptions are rarely justifiable and the computed result can often be found undesirable. Particularly, the simple colour identification technique fails when temperatures of reference leaves, which are embedded in the canopy and provide known conductance to water vapour, are required to be estimated. In this paper, we address this issue and propose a novel algorithm to solve the problem. The underlying plant leaf temperature distribution is considered to be the fusion of two temperature densities separable via a combination of colour identification and Gaussian mixture distribution extraction techniques. A N-average method is tested with moderate success to estimate reference leaf temperatures from the estimated leaf temperature distribution. Our experimental results demonstrate the effectiveness and consistency of the proposed algorithm.