Polar harmonic transforms (PHTs) are a kind of continuous orthogonal moments (COMs) that have excellent image description capability. To improve the anti-noise and reconstruction performance of PHTs, integer-order PHTs are generalized to order by a set of novel decimal-order polar harmonic transforms (DPHTs). On this basis, in combination with trinion theory, the DPHTs are extended to trinion decimal-order polar harmonic transforms (TDPHTs) applicable to color images. The color image is processed as a whole, and the internal relations among three components of color images are fully reserved and utilized. Compared with quaternion color image processing, trinion color image processing effectively avoids information redundancy and improves computational efficiency. Color image reconstruction and zero-watermarking algorithm experiments show that TDPHTs have excellent color image description ability and robustness.