Manual cephalogram marking has a long way from marking on tracing sheet to the availability of commercial softwares for cephalometric analysis. With the effort involved in manual marking and time consumption, it becomes imperative for modern science to envisage algorithms which could automatically locate landmarks on the cephalogram images and perform the various analysis. In this work, we hereby propose a wavelet transform based feature extraction algorithm for detection of the landmark on cephalogram images. 15 landmarks were detected on the images using wavelet transform and all landmarks were detected within the acceptable accuracy limits. This algorithm may have a promising approach in detection of further anatomical landmarks automatically and analysis and thus may help orthodontic practitioners in better and faster treatment planning.