Research On The Key Techniques Of Image Recognition Automatically Of Date Pests

Jujubes have the largest planting area in Chinese drying fruit trees, they are also the important medicinal plants and the ecological and economic tree species. With rapid expansion of the jujube cultivation area, the occurrence of pests and diseases on jujube has become serious year by year, and the hazards bring enormous economic losses to the farmers who plant jujubes. Therefore, the research on the technologies of image recognition automatically of date pests is necessary and possible. It will be conducive to China's characteristics jujube industry information technology and modernization and also has a practical significance for promoting local economic development. In this paper, the image automatic recognition of date pests is researched, which involved digital image processing and image recognition methodsIn this paper, an effective color image segmentation method based on the two-dimensional histogram of the HSI color space is proposed. First, compared with a variety of image preprocessing method, the histogram equalization technology is applied to the image enhancement and the median filter technology is applied to the image filtering of jujube pest images. These image preprocessing methods improved the image dynamic range and contrast and made the image detail outstanding and easy-to-separated. Then the color image segmentation method based on the two-dimensional histogram of the HSI color space is applied to the preprocessed images.In the image feature extraction, the color, texture and shape characteristics of the jujube pests images were analyzed respectively. Totally 20 date pest image features are extracted, such as Color moments, gray level co moments and Hu moments. Finally, in the pattern matching stage, a classifier based on BP neural network is designed.The results show that using the proposed image segmentation method the pest image can be cut up from the complex background and the feature vectors composed of 20 date pest image features is effective for identifying the date pests.