A method for parallel processing of color components of a video signal and an apparatus therefor are disclosed. Specifically, there is provided a method for decoding images based on an intra prediction mode, the method comprising: parsing a syntax element indicating whether a CCLM (cross-component linear model) intra prediction on a current chroma block is disabled or not; deriving an intra prediction mode applied to the current chroma block; and generating a prediction sample for the current chroma block based on the intra prediction mode, wherein, when the CCLM intra prediction is disabled, the intra prediction mode is derived in a predetermined first prediction mode set, when the CCLM intra prediction is not disabled, the intra prediction mode is derived in a predetermined second prediction mode set, and the second prediction mode set includes intra prediction modes included in the first prediction mode set and the CCLM intra prediction mode.