We consider tracking algorithms for manoeuvring targets when the observations include extra information on the current operating mode of the target obtained from an image sensor. The target is modelled as a Markov jump linear system and the image-based observations form a discrete-time point process. We derive the optimal (minimum mean square error) filtered estimate which intrinsically fuses the image-based and primary observations. This optimal filter is computationally prohibitive but provides the basis for a clear understanding of various suboptimal approaches. We propose the image-enhanced IMM filter as a practical alternative which retains many desirable properties of the optimal filter and outperforms existing image-enhanced tracking algorithms over a broad range of operating scenarios.