The present disclosure relates to a processing method of a video signal, the processing method comprising the steps of: scaling a transform coefficient for a current block on the basis of an intermediate scaling factor array; when the flag indicates that a low frequency non-separable transform is applied to the current block, obtaining a residual for the current block by applying an inverse transform of a non-separable transform and an inverse transform of a primary transform on the scaled transform coefficient, wherein the primary transform is a transform applied to a residual signal of a spatial domain before the low frequency non-separable transform; and reconstructing the current block on the basis of the residual and a predictor of the current block.