A Sensor is a device that reads the attribute and changes it into a signal that can be simply examined by an observer or instrument. Sensors are worked in daily objects like touch-sensitive elevator buttons, road traffic monitoring system and so on. Each sensor would carry distinctive capabilities to utilize. The objects obtained in the sensor are tracked by many techniques which have been presented earlier. The techniques which make use of the information from diverse sensors normally termed as data fusion. The previous work defined the object tracking using Multi-Phase Joint Segmentation-Registration (MP JSR) technique for layered images. The downside of the previous work is that the MP JSR technique cannot be applied to the natural objects and the segmentation of the object is also being an inefficient one. To overcome the issues, here we are going to present an efficient joint motion segmentation and registration framework with integrated layer-based and feature-based motion estimation for precise data fusion in real image sequences and tracking of interested objects. Interested points are segmented with vector filtering using random samples of motion frames to derive candidate regions. The experimental evaluation is conducted with real image sequences samples to evaluate the effectiveness of data fusion using integrated layer and feature based image segmentation and registration of motion frames in terms of inter frame prediction, image layers, image clarity.