Pixel selective wavelet transform methods
A method of transforming and sending data, comprising: receiving an image with an associated mask; transforming the received image; quantizing the transformed image; encoding the quantized image and associated mask; and transmitting the encoded image and mask; wherein transforming the received image comprises transforming the dimension subset by repeating steps of: preparing an input vector; determining a filter type; determining a first intermediate vector; determining a second intermediate vector; computing a transform vector from the sum of the first intermediate vector and the second intermediate vector; and computing an output value from the input vector and the computed transform vector.
1. A method of executing a wavelet transform on masked data, the method executed under the control of a processor configured with computer executable instructions, the method comprising:
shuffling, by a system having one or more processors, an unmasked indication associated with a focus input value of a high-pass wavelet filter, to a location in a wavelet filter output mask layer associated with a result of a low-pass wavelet filter; and
encoding, by the system, mask information of the masked data and quantized unmasked results of the wavelet transform.
2. The method of claim 1 , wherein the mask information comprises an initial mask.
3. The method of claim 1 , wherein:
a data value of the masked data initially associated with a focus value of the low-pass wavelet filter is initially masked; and
an output of the encoding the quantized unmasked results of the wavelet transform is independent of the data value.
4. The method of claim 1 , wherein as a consequence of repeated shuffling operations, each unmasked data value of the masked data contributes to a final level wavelet decomposition low-pass filter result.
5. The method of claim 1 , further comprising constructing, by the system, a weight vector for a decomposition filter in the wavelet transform by assembling two partial weight vectors, each of the two partial weight vector determined by a different portion of an input vector mask of the decomposition filter.
6. The method of claim 1 , wherein the shuffling comprises exchanging the unmasked indication and a first data value associated with the focus input value of the high-pass wavelet filter with a masked indication and a second data value associated with a focus value of the low-pass wavelet filter.
7. The method of claim 1 , further comprising repeating, by the system, the shuffling in combination with wavelet filtering until the wavelet transform is complete.
8. The method of claim 1 , wherein the shuffling comprises masking a result of the high-pass wavelet filter, the high-pass wavelet filter associated with an unmasked focus input value, in combination with unmasking the result of the low-pass wavelet filter, the low-pass wavelet filter associated with a masked focus input value.
9. A method of decoding wavelet transformed masked data, the method executed under the control of a processor configured with computer executable instructions, the method comprising:
decoding, by a system having one or more processors, the wavelet transformed masked data to generate wavelet coefficients and mask information;
determining, by the system, from the mask information, a set of mask layers and a set of associated shuffling operations, each of the set of mask layers associated with a level of wavelet transformation;
arranging, by the system, the wavelet coefficients based on the set of mask layers;
inverse wavelet transforming, by the system, each layer of the wavelet coefficients to generate layers of wavelet coefficients;
shuffling, by the system, each layer of the layers of wavelet coefficients based on the set of mask layers and the set of associated shuffling operations.
10. The method of claim 9 , wherein the mask information comprises a mask identifying a placement of unmasked data values resulting from the decoding the wavelet coefficients.
11. The method of claim 9 , wherein one from the set of shuffling operations comprises moving a data value result of an inverse wavelet filter to an adjacent data value location in a layer of the inverse wavelet transforming.