Abstract: A recognition technique of mature pomegranates in natural light conditions was studied．Images of mature pomerganates were enhanced by homomorphic filtering and converted from RGB space to HSI space, then the components of
hue and saturation were fused to remove the dark backgrounds．Afterwards．images were segmented by the regional areas and morphologic characteristics，then centroids and radius were extracted by Hough transform，and the shape of occlusion fruits was recovered．Finally，image processing program was developed for the recognition of mature pomegranates based on VC++6．0．The results showed that the correct recognition rate is>85％，supplying a good foundation for the implementation in the autonomous harvesting on fruit picking in the future．

Key words: homomorphic filter, fusion, hough transform, feature extracting