Complex type seed variety identification and recognition using optimized image processing techniques

H. Salome Hemachitra, A.V. Seetha Lakshmi
2020 ACCENTS Transactions on Image Processing and Computer Vision  
Image processing has been functional to the numerous expansions of agricultural engineering in regulate to accomplish a quick accurate process. The procedure of physical categorization is leisurely and attains a level of bias, which is hard to be enumerated for typical type seed varieties. Seed examination and categorization can afford extra acquaintance in their creation, seeds superiority control and adulteration identification. Several techniques are utilized to resolve the struggle in
more » ... e struggle in perceiving and recognizing the standard type seed varieties, but the most objective is to categorize and recognize the Multifaceted Type Seed Varieties may be a quite difficult process, owing to its textural, shape and color patterns. These techniques do not provide an optimized and a correct depiction of the complex type seed varieties. The main objective of this work is to identify the complex type seed varieties for prospect fertilization within the field of agriculture. This context plans novel image processing systems to recognize, which incorporates an enhanced feature selection, and classification methodologies, which might optimize the exactness and reduce the time consumption of identifying the multifaceted type seed varieties. This novel technique provides efficient identification by feature selection and classification of those composite type seeds. The identification process, Adaptive Median Filter is employed for image enhancement; the edge detection for the image employs Sobel operator and Watershed Segmentation is used for the segmentation. Then Ant Colony Optimization (ACO) strategy is employed for the feature selection and Support Vector Machine ( SVM) is employed in the classification process. The ACO based feature selection (ACOFS) provides ranges 8s to 20s of feature selection time for the dataset and the SVM classification provide 93.487% of accuracy while prediction.
doi:10.19101/tipcv.2020.618022 fatcat:fhjuicqo5jattlrhzzgyz2wrmu