Segmentation Method of Lingwu Long Jujubes based on L*a*b* Color Space

Abstract

Collecting Lingwu long jujubes’ growth characteristics and identifying maturity compared with growth model are important factors in automatic image monitoring system for fruit maturity. To segment and extract regions of Lingwu long jujubes’ against nature scenes in digital images is the key to image recognition. Color space transformation, is used to transform RGB color space to CIE L*a*b* color space, analyze characteristics of three-dimension histogram of each component and increase target display properties by using each component’s point operation. Also, the image pixel threshold segmentation method of each component is proposed; then the target image can be obtained by mathematical morphological and logical operation of the extracted image. 50 images of Lingwu long Jujubes are utilized to compare segmentation results of algorithm with those of manual segmentation method. Comparison results show that the proposed segmentation algorithm can be effectively applied to extracting Lingwu long jujubes’ images against nature scenes, and its accuracy has reached 92.6%.This algorithm has decreased the segmentation error caused by image noise while extracting targets which has strong robustness to the color distribution fluctuation as well.

DOI: http://dx.doi.org/10.11591/telkomnika.v11i9.3289