Research On Core Image Compression And Its Database Establishment

The effect of core image is very important in oil and gas exploration. Because the amount of core images collected is very huge, the compression is necessary for core image storage and transmission. Usually core images have rich texture features and complex edginess. Traditional compression algorithms based on wavelet transform have poor direction selectivity, so it is not conducive to save image details. Therefore, this dissertation makes a combination of Contourlet and wavelet transform, and proposes a reasonable algorithm to compress core image, which could improve the compression effect and visual quality. An image database is established to store the core image compressed and retrieve based on the texture features of it. The main research contents of this paper are as follows:(1) A core image compression algorithm using visual texture degree based on WBCT is put forward. Based on WBCT, this dissertation proposes a concept of Image Texture Measure(ITM) according to the texture feature of core image which is regarded as metrics of direction decomposition number. Combining HVS, this transformation can achieve the optimal decomposition of Contourlet direction transform, and then use SPECK algorithm to encode core image. The experimental results show that, the compression effect of this algorithm proposed is better than that of SPECK obviously, the PSNR is higher and the image reconstructed accords with the features of human vision.(2) A core image compression algorithm using tile technology based on WBCT is put forward. Taking into account the situation that most of core images collected are irregular, this dissertation combines the tile technology in map service with SPIHT algorithm to divide and coed core images. The experimental results show that the reconstructed images have good visual quality, the PSNR value is higher and the texture features of core images are preserved effectively.(3) An image database is established to store the core images compressed. This dissertation proposes a method for extracting core image eigenvalues with WBCT and ITM, which could achieve retrieval of core image in the WBCT domain with Standardized Euclidean Distance(SED) metric. The results prove the algorithm proposed is feasible and effective, the core images retrieved meet the demand and the retrieval results are more accurate.