Texture Segmentation Based on Wavelet Scattering Convolution Network

For image segmentation by texture features of the object,this paper presents a texture segmentation method based on scattering wavelet convolution network.First of all,using scattering wavelet convolution network to extract the scattering energy distribution features of each sub-images,and then using the Ostu global threshold processing method to implement the classification of scattering energy feature matrix at all levels,the coarse segmentation results are given.Finally,to remove spurious spots in the coarse segmented image,applying the method of morphology,thereby achieving unsupervised texture segmentation.Experimental results indicate that high accuracy can be achieved for both texture segmentation and license plate location with the proposed methods.