This paper introduces video compression and image processing methods for fast multimedia channels. Standards designed for image and video coding (JPEG2000, MPEG-4 and MPEG-7), being not strictly defined, present a large field for further experiments and improvements. Our method was developed on fundamentals of Vector Quantization (VQ) and the most advantageous qualities of Karhunen-Loeve Transform (KLT) (known also as Principal Component Analysis - PCA). We also present a direct an idea of how to implement optical transform that can be used with accordance to the methods presented. Among other compression techniques, KLT optimality is discussed as a purpose for further improvement to the method.