Yoga has earned more attention over the past decade due to its significant physical and psychological benefits, including stress reduction, improved flexibility, and muscle strength. Even though the rapid advancements in artificial intelligence, particularly in computer vision, have led to a surge in the number of studies on yoga posture recognition, the sector of pose similarity evaluation still needs to be investigated further. This study introduces a new pose-specific similarity assessment method for yoga postures, determining the most optimal features for each posture. The relevant attributes for each pose are identified using an unsupervised feature selection technique known as Principal Component Analysis (PCA). These properties are compiled into a dictionary with the pose index as the key that can be used to compare different poses. This proposed strategy is a substantial improvement compared to conventional generalized methods. It shows more precise results by focusing on the unique characteristics of poses.