Contour Detection Model Based on the Combination of Surround Facilitation and Inhibition

At the single-cell level in the visual system, the properties of the receptive field (RF) are important bases of visual information processing. The area surrounding classical receptive field (CRF) is called the non-classical receptive field (nCRF), the modulatory effects of which are regarded as an important basis of visual information processing including contour detection. The modulation type (facilitation or inhibition) varies in different neurons, among which inhibition plays the role of removing texture from the background and facilitation plays the role of compensating for weak contours. In this study, we combine two modulation types of the nCRF-i.e., facilitation and inhibition- in a new way to build a contour detection model (denoted by CFI) for cluttered scenes. Inspired by the neuron linking mechanism in V1, the facilitation method is defined based on neuron linking properties and stimulus orientation properties. Our results demonstrate that with the introduction of the facilitation method and cooperation mechanism, more texture edges are removed and more meaningful contours are detected. The proposed model is of great value in contour detection and other advanced computer vision tasks including shape-based object recognition and object segmentation.