The invention relates to a method of image capture in an image sensor with a matrix of rows and columns of active pixels, supplied between a first power supply terminal of zero potential (Vss) and a second power supply terminal at a positive supply potential (Vdd). Each pixel comprises a photodiode and a gate for transferring the photogenerated charges to a charge storage node. The transfer gate is brought by a charge pump to a negative potential (VNEG) for the duration of the charge integration and receives a transfer control signal (TRA) common to all the pixels during a transfer time window. The transfer or reinitialization control signal comprises in succession at least one first phase during which it passes from the negative potential (VNEG) to a positive transfer potential, a second phase during which it passes from the positive transfer potential to the zero potential provided by the first power supply terminal, a third phase of finite duration during which it remains at the zero potential, and a fourth phase in which it passes back from the zero potential to the negative potential provided by the charge pump. The fact of involving the power supply potential limits the inrushes of current into the charge pump which must re-establish the negative potential (VNEG) at the end of the transfer.