Method for the compression of digital images and device for the application of a method of this type

Claim 1

A method for compressing a digital image made up of pixels by assembling said pixels into classes, said method comprising, sequentially testing said pixels as follows, (i) if the distance (d) between the pixel being texted and the nearest identifying value (U i ) of an already existing class is less than a parameter (S), inserting the pixel being tested into the class characterized by the nearest identifying value (U i ), (ii) if the distance (d) between the pixel being tested and the nearest identifying value (U i ) is larger than the parameter S and the number of classes is less than a predetermined number, forming a new class for said pixel, (iii) if the distance (d) between the pixel being tested and the nearest identifying value (U i ) is larger than the parameter S and the number of classes is equal to said predetermined number, incrementing the parameter S and retesting the previously tested pixels.

Claim 2

A method for compressing a digital image made up of pixels by assembling said pixels into classes, said method comprising, sequentially testing said pixels as follows, (i) if the distance (d) between the pixel being texted and the nearest identifying value (U i ) of an already existing class is less than a parameter (S), inserting the pixel being tested into the class characterized by the nearest identifying value (U i ), (ii) if the distance (d) between the pixel being tested and the nearest identifying value (U i ) is larger than the parameter S and the number of classes is less than a predetermined number, forming a new class for said pixel, (iii) if the distance (d) between the pixel being tested and the nearest identifying value (U i ) is larger than the parameter S and the number of classes is equal to said predetermined number, incrementing the parameter S and retesting the previously tested pixels.

Claim 6

A method for compressing a digital image comprising pixels by forming up to 2 K classes, each of said classes having a radiometric identifying value U i , said method comprising (1) setting a parameter S to zero (2) reading a first pixel having radiometric position P 11 (3) forming a first class having a radiometric identifying value U i set equal to P 11 (4) reading an additional pixel having radiometric position P mn (5) calculating the distance d between the nearest radiometric identifying value U i of an already existing class and comparing the distance d with the parameter S (6) if d is equal to or less than S, inserting the pixel into the class having the identifying value U i and reclaculating the value U i , (7) if d is greater than S, determining if the number of classes already formed is less than or equal to 2 K , (8) if the number of classes formed is smaller than 2 K , forming a new class whose identifying value U i is set equal to the value P mn of the last read pixel, and the operation is repeated starting from step (4) above unless the last read pixel is the last pixel of the image in which case the radiometric values of the pixels and the radiometric identifying values of the classes are stored, (9) if the number of classes already formed is equal to 2 K no new class is created, the parameter S is incremented and the process is repeated starting from step (2) above.

Claim 7

A method for compressing a digital image comprising pixels by forming up to 2 K classes, each of said classes having a radiometric identifying value U i , said method comprising (1) setting a parameter S to zero (2) reading a first pixel having radiometric position P 11 (3) forming a first class having a radiometric identifying value U i set equal to P 11 (4) reading an additional pixel having radiometric position P mn (5) calculating the distance d between the nearest radiometric identifying value U i of an already existing class and comparing the distance d with the parameter S (6) if d is equal to or less than S, inserting the pixel into the class having the identifying value U i and reclaculating the value U i , (7) if d is greater than S, determining if the number of classes already formed is less than or equal to 2 K , (8) if the number of classes formed is smaller than 2 K , forming a new class whose identifying value U i is set equal to the value P mn of the last read pixel, and the operation is repeated starting from step (4) above unless the last read pixel is the last pixel of the image in which case the radiometric values of the pixels and the radiometric identifying values of the classes are stored, (9) if the number of classes already formed is equal to 2 K no new class is created, the parameter S is incremented and the process is repeated starting from step (2) above.

Claim 8

A device for compressing a digital image composed of N×M pixels, each of said pixels being represented by B bits, said device comprising a processing unit for receiving said digital image signal and for delivering a compressed digital image signal comprising pixels encoded in K bits and a digital signal corresponding to the encoding of 2 K classes in B bits, an image memory with a volume of M×N×K bits in which said compresed digital image is stored, a transcoding memory having a volume of 2 K ×B bits for storing the digital signal corresponding to the encoding of said classes, said transcoding memory being adapted to receive a signal comprising M×N×K bits from said image memory and to transmit a signal comprising M×N×B bits, an analog-to-digital converter connected to said transcoding memory, for converting said M×N×B bit signal from said transcoder memory to analog form from display on a screen, said processor being adapted to compress said digital image by organizing said pixels ito said 2 K classes by sequentially processing said pixels as follows: (i) if the radiometric distance (d) between the pixel being processed and an identifying value U i of an already existing class is less than a parameter (S), inserting the pixel beig processed into the class characterized by the nearest identifying value, (ii) if the distance (d) between the pixel beig processed and the nearest identifying value (U i ) is larger than the parameter (S) and the number of existing classes is less than 2 K , forming a new class for said pixel, (iii) if the distance (d) between the pixel beig processed and the nearest identifying value (U i ) is larger than the parameter (S) and the number of existing classes is equal to 2 K , incrementing the parameter (S) and retesting the previously tested pixels.

Claim 9

A device for compressing a digital image composed of N×M pixels, each of said pixels being represented by B bits, said device comprising a processing unit for receiving said digital image signal and for delivering a compressed digital image signal comprising pixels encoded in K bits and a digital signal corresponding to the encoding of 2 K classes in B bits, an image memory with a volume of M×N×K bits in which said compresed digital image is stored, a transcoding memory having a volume of 2 K ×B bits for storing the digital signal corresponding to the encoding of said classes, said transcoding memory being adapted to receive a signal comprising M×N×K bits from said image memory and to transmit a signal comprising M×N×B bits, an analog-to-digital converter connected to said transcoding memory, for converting said M×N×B bit signal from said transcoder memory to analog form from display on a screen, said processor being adapted to compress said digital image by organizing said pixels ito said 2 K classes by sequentially processing said pixels as follows: (i) if the radiometric distance (d) between the pixel being processed and an identifying value U i of an already existing class is less than a parameter (S), inserting the pixel beig processed into the class characterized by the nearest identifying value, (ii) if the distance (d) between the pixel beig processed and the nearest identifying value (U i ) is larger than the parameter (S) and the number of existing classes is less than 2 K , forming a new class for said pixel, (iii) if the distance (d) between the pixel beig processed and the nearest identifying value (U i ) is larger than the parameter (S) and the number of existing classes is equal to 2 K , incrementing the parameter (S) and retesting the previously tested pixels.