Image transition region extraction and segmentation method based on wavelet transform

It is very important to segment the complicated tree image precisely in precision toward-target pesticide application and intelligent plant-protection machinery design. The segmentation method based on image transition region extraction was supported in view of tree image features. The wavelet transform feature coefficients which could break down much more high-frequent and low-frequent information were determined by comparing wavelet transform coefficients, coefficient clustering and wavelet box coefficients. Wavelet energy ratio parameter was defined based on wavelet feature coefficients and clustered into image gray value. Both self-adjusting threshold and neural network methods were employed to extract the transition area on which the tree images with complicated background were segmented based. The experiments showed that the images were segmented precisely and the method was superior to other methods in segmenting complicated images.