The Research And Development On Image Of Flowers Segmentation And Processing Based On Region And Active Contour

Facilities flower is the important part of facilities argriculture.It can be used to plant appreciable varieties of flower which demand high environmental factor.At the same time,facilities flower can improve the economic benefit of planting flowers,and it is on the top of the flower industry. The research on segmentation and feature extraction of flower images is the key technology of making facilities flower digital,automatic and intelligent.It can achieve the growth parameters of facilities flower automatically and control the growth period of facilities flower.This paper has research on segmentation and feature extraction of natural image of flower(e.g. phalaenopsis amabilis image). The main research areas are as follows:1. The research history and development status about domestic and foreign technologies including natural image segmentation and feature extraction,especially methods based on region information and active contour were analyzed and summarized.2. Image segmentation methods based on region information and active contour and feature extraction methods were researched and analyzed.On region methods, region growth,regional division and return and watershed were analyzed.On active contour methods,parameter active contour and geometry active contour were analyzed. Color feature,shape feature and texture feature in respect of feature extraction were analyzed.3.On the foundation of research on watershed algorithm,this paper proposed improved watershed algorithm which is used to segment natural image background. By using gradient information extract significant edge and clearing the unconspicuous "dam", it inhibit the watershed alogrithm over-segmentation phenomena. So the effect of the algorithm for natural image segmentation is better.4.Research on active contour model and morphological skeleton algorithm. Improved active contour model was proposed,which was used to extract the single flower from flower cluster.In active contour model,through improved skeleton algorithm and contour repossession algorithm,the initial iteration profile could be achieved.Then,the frame generated by skeleton algorithm was used to form a shape energy term as the external force constraints of active contour model.Improved active contour model increased the segmentation accuracy,reduced the sum of iterations and increased the speed of image processing.5.Feature information about phalaenopsis amabilis,including color feature,texture feature and shape feature were researched.Through the white regions in the center of flower and improved active contour model,extraction of single flower was realized.H channel of HIS color space was used to extract the color feature of flower.Relevant image processing methods were used for acquiring area,perimeter,circularity,coordinate of center,lengthwidth ratio and other shape parameters.At last,a software about segmentation and texture extraction of natural flower image was developed,and it was used to test the above-mentioned segmentation and texture extraction methods.