Abstract:  In this study， a threshold\|based algorithm was designed for image segmentation of bananas in the crate. The pixels of banana and background were selected discretely in the manual\|segmented regions. Subsequently， nine color features were extracted from RGB， HSV and CIE L*a*b* color spaces. Three thresholds from B， L* and b* color channels were used for the design of the algorithm， which were determined by the statistical analysis. The qualitative assessment of segmentation performance was evaluated by the application of intuitive comparison， and the obtained results showed that the outlines of manual\|segmented and automatic\|segmented regions were almost similar. Simultaneously， the quantitative evaluation of segmentation performance was carried out by area ratio， and the average area ratio of ten tested samples was beyond 80%. These two tested results indicated that the performance of this automatic segmentation algorithm might be satisfying， and this methodology could provide the key technology for the design and achievement of the real\|time quality monitoring and control system in banana ripening rooms.

Key words: banana, image segmentation, computer vision, threshold technique, color space