Image processing is traditionally a very compute-intensive process. In case of stringent timing constraints, traditional approaches scale down application quality, therefore, compromising the visual clarity of processed images. In order to overcome such drawbacks, we follow a new resource-aware parallel computing paradigm called invasive computing in this paper, where an application can dynamically claim, execute, and release resources on a multi-core computing system. In this context, we show how an invasive image processing application is able to make run-time tradeoffs of quality or throughput depending on the requirements and the number of available processing resources. As a target architecture, a class of massively parallel architectures called tightly-coupled processor arrays is chosen, to show the adaptivity provided by invasive computing. The applications gain the ability to fulfill constraints in two directions: a) In case of strict throughput requirements, the image quality may be adjusted in dependence on the number of available resources by run-time selection and loading of different algorithmic kernels, e.g., image filters in dependence of the success of invasion. b) Alternatively, a certain level of quality can be guaranteed by dynamically adjusting the throughput with respect to available resources.