Adaptive Color Transform Signaling in Video Coding

The patent changes the parameter of ACT flag signaling from a fixed always-on approach to a conditional approach based on the presence of coefficients in the CU block. The encoder signals the ACT flag only when coefficients are present, and the decoder infers the flag value based on the absence of signaling, thereby eliminating redundant signaling overhead while maintaining complete coverage of all cases.

Instead of signaling the ACT flag for all CU blocks (excessive action), the patent applies partial signaling only where necessary - specifically when the CU block contains coefficients. This partial action approach reduces redundant signaling while ensuring ACT mode is properly indicated for all relevant blocks through a combination of explicit signaling and implicit inference.

The patent creates a universal constraint handling mechanism that applies to both intra and inter blocks. The rule that 'if the ACT flag is not signaled, then the CU coded flag must be 1' serves both prediction modes uniformly, eliminating the need for separate constraint handling logic for intra and inter blocks while maintaining reliability of ACT mode indication.

Instead of specifying different constraints for intra and inter blocks (positive specification), the patent inverts the approach by specifying a default constraint that applies to both, and only allowing exceptions when explicitly signaled. This inversion simplifies the constraint system while maintaining the necessary distinctions between prediction modes.