COMPRESSION OF HYPERSPECTRAL IMAGES USING WATERSHED METHOD AND WAVELET TRANSFORM

One of the first processes that is proposed about hyperspectral data due to the increased number of imaging bands and the possibility of increasing the radiometric accuracy, is their compression. In this paper, a method is provided for compressing the hyperspectral images. In the proposed compression method, at first compress the image using the wavelet transform method that is a spectral compression method, then the segmentation is performed on the image using watershed method, so that the image is divided into areas with pixels of similar properties. Considering the high volume of data, the hyperspectral image can be sent rather than the segmented hyperspectral image along with the estimated parameters such as mean, variance, seed pixel and texture block of any area that are obtained from the extraction of spectral feature of wavelet transform and in the receiver side, the image can be reconstructed using the transmitted information. This method, considering that it performs the compression using the image segmentation, identifies the small areas which could be removed in the other compression methods, and maintains them during the compression. This method has been implemented on AVIRIS image and created a high compression ratio and also maintained the information about classification.

Keywords:

Hyperspectral Images , Watershed , Segmentation , Compression