Abstract:
Images are everywhere and play significant role in our everyday life and scientific research. Captured images may not be in good quality because of some undesirable conditions such as bad lighting, and capturing devices itself etc. Contrast is an important factor of image quality. Human eye and certain applications are very sensitive to spatially varying contrast but may not identify information with little variation. In this thesis, parametric and non-parametric contrast enhancement techniques are proposed
for grayscale and color images. The proposed non-parametric technique “histogram equalization using weighted transformation function for contrast enhancement (CEWTF)” utilizes the concept of iterative spatial size reduction, smoothing and weighting histogram, and transformation function weighting. Parametric technique which controls degree of enhancement uses the concept of power law transformation. For color image contrast enhancement, proposed technique is based on HSV color model. It only redistributes luminance component while preserving hue and saturation. Subjective comparison with other state of the artmethods is performed by considering improved contrast with naturalness, colorfulness and absence of unwanted artifacts. Enhanced images are also analysed through quantitative measures.
Simulation results on different images show the significance of proposed technique as
compared to state-of-the-art existing techniques.
