A low frequency component image L[i] and high frequency component-emphasized image SH[i] are generated from an image A[i]. Lossy compression processing is performed for the low frequency component image L[i] to generate a compressed image C[i] and store it in a memory. A compressed image C[i−1] is decoded to generate a decoded image L′[i−1]. The compressed image C[i] is decoded to generate a decoded image L′[i]. A difference image E[i] between the decoded image L′[i] and the low frequency component image L[i] is generated. The low frequency component image L[i], decoded image L′[i−1], and difference image E[i] are composited at a predetermined ratio to generate a composite image SL[i]. The high frequency component-emphasized image SH[i] and composite image SL[i] are output as subframe images of the i-th frame.