Value Representations (VR) are format or type definitions for DICOM elements (Attributes). The VR for each data element specifies the encoding of the data. For example, for all person names, we need an agreement on whether one should specify first the family name or the first name.
The specific VR is specified for each Attribute in the DICOM data dictionary. The individual attributes listed in the data dictionary are identified by tags, which are codes consisting of two hexadecimal components, aka Group and Element numbers. For example, the patient name has a tag of (0010, 0010).
The data dictionary lists all attributes and VRs. For example, it defines that all person names in a DICOM message (i.e. Patient Name, Referring Physician Name, Ordering Physician Name, etc.) have the VR identified with the two-character PN (Person Name). The format of each of the standard VRs is given, including a list of legal characters and any delimiters that one might use.
When exchanging an object such as an image, the VR for all elements in the message can be implicitly or explicitly encoded. Which method to use is negotiated between two systems during the association set-up. Explicit VR requires that the transmitted data will include the VR information, indicated by its two character identifier, along with data and attribute tags for each individual attribute. Implicit VR requires that the transmitted date will not include the VR information. The receiving application must in this case determine the VR type from the attribute tag using an internal data dictionary, meaning when receiving a tag (0010,0010) it should know this is encoded using the PN convention. In the case of an implicit value representation, this is assumed to be known by the receiver and not explicitly specified