Several competitive techniques have been applied for efficient image segmentation and automatic feature extraction through the literatures. There are a lot of open problems and controversial ambiguities regarding to the mechanism which applied by human eye for image segmentation and feature extraction. Here we have first extracted the human vision technique applied for image segmentation and we have implemented this technique for automatic image segmentation and feature extraction. The features have been categorized into the internal and external modalities. We have introduced the negative curvature minima (NCM) points as a dominant external feature and the textures detected using pulse coupled neural networks (PCNNs) and LAWs methods as the dominant internal feature used by human vision to segment and extracts the features of an image. These features have been used to detect suspicious masses in mammogram images using the proposed human eye inspired technique. The results justify the efficiency of the proposed method.