Homomorphic Filtering of colour images using a Spatial Filter Kernel in the HSI colour space

This paper describes a solution to the issue of providing a balance between reduced computational complexity and image quality for homomorphic filtering of colour images. The solution eliminates frequency-domain transforms and calculation of the filter kernels dependent on image size in the frequency domain. It also reduces the overall computational complexity associated with frequency domain filtering and is easier to implement in hardware for real-time applications. The solution also reduces demands on hardware memory by processing the image in the HSI colour space. Additionally, the HSI domain helps reduce computational complexity further since only one channel is processed. Furthermore, the spatial domain filter kernel provides excellent contrast enhancement and sharper edges while the HSI colour space processing gives good colour rendition, thus improving colour image quality and removing the need for colour restoration postprocessing.