Flower Image Recognization Based On Two-dimensional Wavelet

The cornerstone of the wavelet transform is the Fourier transform.The Fourier transform is affected by the uncertainty principle or how the Fourier transform lacks resolution between the frequency domain and the time domain.Decomposing the signal into wavelets rather than frequencies can give better resolution in the converted domain However,when the wavelet transform is used,the signal is converted to the wavelet domain rather than the frequency domain.Wavelet transform can be used in machine learning and image recognition.Machine learning,image recognition and other technologies are widely used in various fields.In domestic and foreign researches,wavelet is applied in image recognition,face recognition and other recognition methods.Generally,there are three or two methods as follows:use wavelet to layer or cut images and then analyze some images.Several scaling functions are used to extract the one-dimensional wavelet coefficients in different directions.The image is reduced to ultra-small dimensions and recognized by two-dimensional wavelet.In the research on the application of two-dimensional wavelet in the field of image recognition,most schemes use the image library,but once the image in the non-image library appears,the recognition rate is very low,in view of the problem of image recognition,it has broad prospects and practical value to research a set of algorithms suitable for more images In this paper,instead of using the image library,single flower pictures are selected.The hierarchical scale coefficient of two-dimensional wavelet is extracted,and the dimensionality is reduced by principal component analysis.The characteristic values that can identify flower species are selected.The main research work of this paper is as follows:based on the basic processing of image,the gray matrix of image is extracted,and the hierarchical approximate wavelet coefficient of image is extracted.Based on principal component analysis,the wavelet coefficients are reduced.Based on the decision tree,the coefficients were trained and predicted,and the decision tree was bagged 10,000 times to obtain the correct rate of flower image recognition of this model.