Compression and decompression of image data using minimized size matrix algorithm

摘要

A computer-implemented method for compressing image data (104, 200) comprising: applying (210) a discrete cosine transform (DCT) to each of a plurality of non-overlapping pixel blocks spanning a frame of image data to generate a set of DCT coefficients for each block of pixels comprising a DC DCT coefficient and a plurality of AC DCT coefficients; quantizing (230) each set of DCT coefficients to generate a set of quantized DC DCT coefficients and a set of quantized AC DCT coefficients; forming (240) a DC matrix from the set of quantized DC DCT coefficients; forming (250) an AC matrix from the set of quantized AC DCT coefficients; forming (352) a bounded data array comprising elements that have values corresponding only to each unique value of the elements in array AC; compress (350) matrix AC by removing data blocks from matrix AC that have only zero values, where each data block has the same known fixed size, and forming a reduced AC matrix from data blocks in the matrix AC that includes zero values; storing the position in matrix AC of each data block in matrix AC, including non-zero values in a location matrix; generating a key (402) using the maximum value of the elements of the reduced AC array; compress (368, 400) the reduced AC matrix using the key to form an encoded AC matrix, where the key comprises a plurality of key components and the same number of elements of the reduced AC matrix as the number of key components are combined using the key to form a single element of the coded AC matrix; arithmetically encoding (270) the DC matrix and the encoded AC matrix to form arithmetically encoded data; and forming a compressed image file (430), which includes the arithmetically encoded data (436, 438, 440) and stores the location matrix in a header of the compressed image file, and stores the key (432) and the matrix of limited data (434).