Image segmentation method by inosculating color factor and spatial information for eggplant

An algorithm for eggplant image is brought forward by merging the G-B color factor and spatial information. In order to find the optimum color space and color mode for image segmentation, the grayscale and histogram of 40 picture under different light intensities are extracted. The characters of their fruits, leaves, stems and void are studied and the G-B color mode is most available to segment the image of eggplants. The seed regions are determined according to both uniform gray-level and 8 epsilon neighborhood intercommunication, and then the starting segmentation is carried out by scanning image. The initial regions are merged based on the G-B color mode and spatial information. The region growing stop until the regional class distance of these segmented regions is maximum. The residua are got rid of preferably by template operation and morphologic operation. Experiment results show that this segment method with the segmentation efficiency of 92% have higher intelligence and the time consumed is 0.36s and has high intelligent and suitability.