In response to the growing demand of rice farming worldwide, the crop’s increased water demand and the existence of laborious and destructive methods to estimate physiological responses to water stress, we present a novel and alternative method for estimating the crop water stress index (CWSI) in a rapid and non-invasive way through the use of thermal images obtained from the canopy of the crop. This research was implemented in the experimental irrigation area at the La Molina National Agrarian University in Lima, Peru, and the objective was to optimize water use through the relationship between the CWSI and soil moisture, controlling the latter by a drip irrigation system. Given the water needs of the crop, the water potential remained close to zero (T0, control), with decreases to −10 (T1), −15 (T2), and −20 (T3) Kpa, to estimate the relationship between the CWSI and moisture during the vegetative and reproductive stages of the crop. The study was divided into three phases: First, algorithms were developed to extract thermal images only with pixels of vegetation cover. Second, the dry and wet temperatures (Td and Tw), variables required for the calculation, were obtained based on pots subjected to permanent wilting point and saturation, respectively. Finally, the CWSI was associated with the volumetric soil moisture at depths of 12 and 20 cm, achieving the highest correlation at 20 cm (R = –0.489), as a result of the minor effect of soil evaporation. This study allowed the identification of an alternative and innovative method to estimate the rice-crop water stress index in relation to soil moisture, with the possibility of its future implementation in the irrigation schedule.