The heat transfer between the ocean and the atmosphere is one of the most important parameters governing the global climate. Important parameters include the heat transfer velocity and the net heat flux as well as parameters of the underlying transport model. However, the net heat flux is hard to measure since processes take place in the thermal boundary layer, that is the topmost layer of the ocean less than 1 mm thick. Current techniques rely on three independent measurements of the constituent fluxes, the sensible heat flux, latent heat flux and radiative flux. They depend on indirect measurements of meteorological parameters and rely on a combination of data from different sensors using a number of heuristic assumptions. High relative errors and the need for long temporal averaging reduce the practicability of these techniques. In this paper a novel technique is presented that circumvents these drawbacks by directly measuring the net heat flux across the air-water interface with a single low-NETD infrared camera. A newly developed digital image processing technique allows to simultaneously estimating the surface velocity field and parameters of the temporal temperature change. In particular, this technique allows estimating the total derivative of the temperature with respect to time from a sequence of infrared images, together with error bounds on the estimates. This derivative can be used to compute the heat flux density and the heat transfer velocity, as well as the probability density function of the underlying surface renewal model. It is also possible to estimate the bulk-skin temperature difference given rise to by the net heat flux. Our technique has been successfully used in both laboratory measurements in the Heidelberg Aeolotron, as well as in field measurements in the equatorial pacific during the NOAA GasExII experiment this spring. The data show that heat flux measurements to an accuracy of better than 5% on a time scale of seconds are feasible.