Abstract: A system and method is disclosed for compressing and decompressing composite video signals. In one type of operation a composite video signal (e.g., an NTSC signal) is sampled and the samples compressed into 3 or 5 bit composite values, depending on the system mode (PREVIEW or HIGHRES). For the HIGHRES mode the compressed composite values are written into a RAM for at least the four fields associated with the most recent two frames. As needed, the compressed values are transferred to a decoding engine implemented in software. The decoding engine, without any need for phase locked loop circuitry, decompresses the data, and decodes the decompressed digital composite values into YUV and/or RGB values. When the system is in the 5-bit mode, the decoding engine scans and compares all four fields to detect the existence of motion in the image, which could blur the captured image. In regions with motion only a single field of data is used as the source of the image data for those regions.