Apparatus and method for emphasizing a selected region in the compressed representation of an image
AssigneeDavid Sarnoff Research Center, Inc.
InventorJerome M. Shapiro
Life of this patent
Estimating the term…
Abstract
In certain image analysis applications, it is desirable to compress the image while emphasizing a selected region of the image. The invention is a means for to allocating more bits, and thus better quality in the decoded image, to the selected region at the expense of other regions of the image. This allows efficient compression of the image for storage or transmission with those regions deemed to be important preserved at high quality and other regions stored with minimal quality to preserve the context of the image.
Claims12
In a data-compression method for image information defined by the respective amplitudes of pixels comprising an image having given dimensions, wherein higher-amplitude pixel data is considered to be more significant than lower-amplitude pixel data;
- the improvement for emphasizing image information defining at least one region of said image having a specified spatial identity with respect to said given dimensions;
- said method comprising the steps of:
- a) selecting the pixels of said one region to be emphasized;
- b) multiplying the respective amplitudes of the pixels of said one selected region by a scaling factor having a given value larger than unity to effect an increase in pixel amplitude in accordance with said given value of said scaling factor;
- c) following step (b), forming a wavelet transform of the pixel amplitudes for the entire image to derive wavelet coefficients at each of a number of wavelet scales, said wavelet coefficients having values determined by the amplitudes of the image pixels that are transformed;
- d) employing a given bit budget for forming from said wavelet coefficients an encoded data bit stream made up of no more than a given maximum number of bits determined by said given bit budget, wherein a bit of its bit budget is allocated to each and every wavelet coefficient having a value larger than a first given threshold value before a bit of its bit budget is allocated to any wavelet coefficient having a value which is not as large as said first given threshold value but is larger than a second given threshold value which is smaller than said first given threshold value, whereby said wavelet coefficients having values determined by relatively higher-amplitude pixels are allocated more bits in said encoded data bit stream than coefficients having values determined by lower-amplitude pixels;
- e) forming a bit stream header that that defines the given value of said scaling factor and defines the identity of said one selected region with respect to said given image dimensions;
- and
- f) appending the encoded data bit stream to the bit stream header.
In data-compression apparatus for image information defined by the respective amplitudes of pixels comprising an image having given dimensions, wherein higher-amplitude pixel data is considered to be more significant than lower-amplitude pixel data;
- the improvement for emphasizing image information defining at least one region of said image having a specified spatial identity with respect to said given dimensions;
- said apparatus comprising:
- means for selecting the pixels of said one region to be emphasized;
- means for multiplying the respective amplitudes of the pixels of said one selected region by a scaling factor having a given value larger than unity to effect an increase in pixel amplitude in accordance with said given value of said scaling factor;
- means for forming a wavelet transform of the pixel amplitudes for the entire image, after the pixel amplitudes of said one selected region have been increased by said means for multiplying, to derive an array of wavelet coefficients at each of a number of wavelet scales, said wavelet coefficients having values determined by the amplitudes of the image pixels that are transformed;
- means having a given bit budget for forming from said wavelet coefficients an encoded data bit stream made up of no more than a given maximum number of bits determined by said given bit budget, wherein a bit of its bit budget is allocated to each and every wavelet coefficient having a value larger than a first given threshold value before a bit of its bit budget is allocated to any wavelet coefficient having a value which is not as large as said first given threshold value but is larger than a second given threshold value which is smaller than said first given threshold value, whereby said wavelet coefficients having values determined by relatively higher-amplitude pixels are allocated more bits in said encoded data bit stream than coefficients having values determined by lower-amplitude pixels;
- means for forming a bit stream header that specifies the given value of said scaling factor, the given image dimensions, the identity of said one selected region, and the number of said wavelet scales;
- and
- means for appending the encoded data bit stream to the bit stream header.
A decoding method for reconstructing a certain image having given dimensions from a bit stream of image-information data comprising a bit stream header having appended thereto a bit stream of pixel-amplitude information that has been encoded by a data-compression code comprising certain coefficients of a number of wavelet scales of a wavelet transform of pixel amplitudes of a modified image having at least one emphasized region having a specified spatial identity with respect to said given dimensions in which the pixel amplitudes thereof have been increased by a scaling factor of a given value with respect to pixel amplitudes of a region of said certain image corresponding to said one emphasized region, and said bit stream header specifies the given value of said scaling factor, the given image dimensions, the identity of said one emphasized region, and the number of said wavelet scales, wherein said bit stream of pixel-amplitude information that has been encoded by a data-compression code includes a first-occurring set comprising those bits representative of pixel amplitudes having values larger than a first given threshold value and may also include a second-occurring set comprising those bits representative of pixel amplitudes having a value not as large as said first given threshold value but larger than a second given threshold which is smaller than said first given threshold;
- said decoding method comprising the steps of:
- a) separating the bit stream header from the data bit stream;
- b) extracting from said bit stream header the given value of said scaling factor, the given image dimensions, the identity of said one emphasized region, and the number of said wavelet scales specified thereby;
- c) decoding each and every one of the bits of the first set of the data bit stream of pixel-amplitude information before decoding any of the bits of the second set of the data bit stream of pixel-amplitude information to form an array of wavelet coefficients at each of said wavelet scales;
- d) performing the inverse wavelet transform to derive the pixel amplitudes of said modified image;
- and
- e) multiplying the pixel amplitudes of said one emphasized region of said modified image by the reciprocal of said given value scaling factor in deriving said reconstructed certain image.
Decoding apparatus for reconstructing a certain image having given dimensions from a bit stream of image-information data comprising a bit stream header having appended thereto a bit stream of pixel-amplitude information that has been encoded by a data-compression code comprising certain coefficients of a number of wavelet scales of a wavelet transform of pixel amplitudes of a modified image having at least one emphasized region in which the pixel amplitudes thereof have been increased by a scaling factor of a given value with respect to pixel amplitudes of a region of said certain image corresponding to said one emphasized region, and said bit stream header specifies the given value of said scaling factor, the given image dimensions, the identity of said one emphasized region, and the number of said wavelet scales, wherein said bit stream of pixel-amplitude information that has been encoded by a data-compression code includes a first-occurring set comprising those bits representative of pixel amplitudes having values larger than a first given threshold value and may also include a second-occurring set comprising those bits representative of pixel amplitudes having a value not as large as said first given threshold value but larger than a second given threshold which is smaller than said first given threshold;;
- said decoding apparatus comprising:
- means for separating the bit stream header from the data bit stream;
- means for extracting from said bit stream header the given value of said scaling factor, the given image dimensions, the identity of said one emphasized region, and the number of said wavelet scales specified thereby;
- means for decoding each and every one of the bits of the first set of the data bit stream of pixel-amplitude information before decoding any of the bits of the bits of the second set of the bit stream of pixel-amplitude information to form an array of wavelet coefficients at each of said wavelet scales;
- means for performing the inverse wavelet transform to derive the pixel amplitudes of said modified image;
- and
- means for multiplying the pixel amplitudes of said one emphasized region of said modified image by the reciprocal of said given value scaling factor in deriving said reconstructed certain image.
Dependencies are inferred from the claim text.