Histogramming of pixel values on a distributed processing system

Claim 3

The method of claim 1 wherein said element value for each said data element is expressed in binary form with the same bit length, a true head indicator being set at said true head element of said data string, step (b) of said method comprising the substeps of: (d) specifying a predetermined one of the bits of said element value as a partitioning bit, said data string being an original string composed of at least one string fragment, each said fragment being composed of a maximal continuous sequence of said data elements having the same value of said partitioning bit; (e) copying at all said data elements the value of said first pointer into a second pointer; (f) determining for all said data elements except said true head element if the immediately preceding element has the same value of said partitioning bit as itself; (g) setting a first tag at all said data elements for which it is determined in step (f) that the immediately preceding element does not have the same partitioning bit value as itself, each said data element having said first tag so set being a fragment head element; (h) setting said first tag and a second tag only at said true head element, each said data element having said first tag set or not set being respectively a tagged or untagged element; (i) getting, for all said untagged elements, the values of said first and second tags and said second pointer from the data element currently identified by said second pointer of said untagged element; (j) substituting, at all said untagged elements, the values of said first and second tags and said second pointer obtained in step (i) from said currently identified element respectively for said first and second tags and said second pointer of said untagged element; (k) repeating steps (i) and (j) until all of said data elements in said original string are tagged; (l) setting said true head indicator at each said fragment head element for which it is determined that said second tag is set at the immediately preceding element in said original string; (m) getting, only for each said fragment head element, the value of said second pointer of said immediately preceding element in said original string and substituting the last recited second pointer value for the value of said first pointer of each said fragment head element, so that two filaments are formed each being defined by said first pointer at each said data element and said true head indicator being set at the data element at the head of each said filament; and (n) specifying as said partitioning bit a different one of the bits of said element value and repeating steps (e) through (m) unless all bits of said element value have been so specified, steps (e) through (m) being repeated with each said filament formed in the last performance of step (m) being said original string, step (b) being complete if all bits of said element value have been specified as said partitioning bit whereupon each said filament formed in the last performance of step (m) is homogeneous.

Claim 4

The method of claim 1, step (c) including the substeps of: (d) identifying a tail element of each said homogeneous filament; (e) assigning a sequential position value to each said data element in each said homogeneous filament, said true head element of each said homogeneous filament being assigned the sequential position value of zero, the sequential position value of each said data element exceeding by one the sequential position value of the data element identified by its first pointer; and (f) incrementing by one the sequential position value of each said tail element, the incremented position value of each said tail element being the number of said data elements composing the homogeneous filament with which said tail element is associated.

Claim 5

The method of claim 1, step (c) including the substeps of: (d) identifying a tail element of each said homogeneous filament; (e) copying said first pointer into a second pointer at each said data element; (f) defining at each said true head element a sequential position value of zero; (g) setting a tag at said true head element; (h) setting a counter value at zero; (i) determining if said tag is set at each said data element in each said homogeneous filament and, if so, proceeding to step (o) and, if not, proceeding to step (j); (j) taking a value 2 i +v as the sequential position value at each said data element for which said tag is not set, each of the last recited data elements being a receiving element, where: i=the current counter value; and v=the sequential position value of a passing element identified by said second pointer of said receiving element; (k) copying at each said receiving element the tag of said passing element; (l) determining, at each said data element for which said tag is not set, the data element currently identified by said second pointer of said determining element and substituting said second pointer of said currently identified data element for said second pointer of said determining element; (m) incrementing by one said counter value; (n) returning to step (i); and (o) incrementing by one said sequential position value of each said tail element, the incremented position value of each said tail element being the number of said data elements composing the homogeneous filament with which said tail element is associated.

Claim 6

The method of claim 2, step (c) including the substeps of: (o) identifying a tail element of each said homogeneous filament; (p) assigning a sequential position value to each said data element in each said homogeneous filament, said true head element of each said homogeneous filament being assigned the sequential position value of zero, the sequential position value of each said data element exceeding by one the sequential position value of the data element identified by its first pointer; and (q) incrementing by one the sequential position value of each said tail element, the incremented position value of each said tail element being the number of said data elements composing the homogeneous filament with which said tail element is associated.

Claim 9

The method of claim 6, step (d) including the substeps of: (e) identifying a tail element of each said homogeneous filament; (f) assigning a sequential position value to each said data element in each said homogeneous filament, said true head element of each said homogeneous filament being assigned the sequential position value of zero, the sequential position value of each said data element exceeding by one the sequential position value of the data element identified by its first pointer; and (g) incrementing by one the sequential position value of each said tail element, the incremented position value of each said tail element being the number of said data elements composing the homogeneous filament with which said tail element is associated.

Claim 10

The method of claim 6, step (d) including the substeps of: (e) identifying a tail element of each said homogeneous filament; (f) copying said first pointer into a second pointer at each said data element: (g) defining at each said true head element a sequential position value of zero; (h) setting a tag at said true head element; (i) setting a counter value at zero; (j) determining if said tag is set at each said data element in each said homogeneous filament and, if so, proceeding to step (p) and, if not, proceeding to step (k); (k) taking a value 2 i +v as the sequential position value at each said data element for which said tag is not set, each of the last recited data elements being a receiving element, where: i=the current counter value; and v=the sequential position value of a passing element identified by said second pointer of said receiving element; (l) copying at each said receiving element the tag of said passing element; (m) determining, at each said data element for which said tag is not set, the data element currently identified by said second pointer of said determining element and substituting said second pointer of said currently identified data element for said second pointer of said determining element; (n) incrementing by one said counter value; (o) returning to step (j); and (p) incrementing by one said sequential position value of each said tail element, the incremented position value of each said tail element being the number of said data elements composing the homogeneous filament with which said tail element is associated.

Claim 11

The method of claim 6 wherein said element value for each said data element is expressed in binary form with the same bit length, a true head indicator being set at said true head element of said data string, step (c) of said method comprising the substeps of: (e) specifying a predetermined one of the bits of said element value as a partitioning bit, said data string being an original string composed of at least one string fragment, each said fragment being composed of a maximal continuous sequence of said data elements having the same value of said partitioning bit; (f) copying at all said data elements the value of said first pointer into a second pointer; (g) determining for all said data elements except said true head element if the immediately preceding element has the same value of said partitioning bit as itself; (h) setting a first tag at all said data elements for which it is determined in step (g) that the immediately preceding element does not have the same partitioning bit value as itself, each said data element having said first tag so set being a fragment head element; (i) setting said first tag and a second tag only at said true head element, each said data element having said first tag set or not set being respectively a tagged or untagged element; (j) getting, for all said untagged elements, the values of said first and second tags and said second pointer from the data element currently identified by said second pointer of said untagged element; (k) substituting, at all said untagged elements, the values of said first and second tags and said second pointer obtained in step (j) from said currently identified element respectively for said first and second tags and said second pointer of said untagged element; (l) repeating steps (j) and (k) until all of said data elements in said original string are tagged; (m) setting said true head indicator at each said fragment head element for which it is determined that said second tag is set at the immediately preceding element in said original string; (n) getting, only for each said fragment head element, the value of said second pointer of said immediately preceding element in said original string and substituting the last recited second pointer value for the value of said first pointer of each said fragment head element, so that two filaments are formed each being defined by said first pointer at each said data element and said true head indicator being set at the data element at the head of each said filament; and (o) specifying as said partitioning bit a different one of the bits of said element value and repeating steps (f) through (n) unless all bits of said element value have been so specified, steps (f) through (n) being repeated with each said filament formed in the last performance of step (n) being said original string, step (c) being complete if all bits of said element value have been specified as said partitioning bit whereupon each said filament formed in the last performance of step (n) is homogeneous.

Claim 12

The method of claim 9, step (d) including the substeps of: (p) identifying a tail element of each said homogeneous filament; (q) assigning a sequential position value to each said data element in each said homogeneous filament, said true head element of each said homogeneous filament being assigned the sequential position value of zero, the sequential position value of each said data element exceeding by one the sequential position value of the data element identified by its first pointer; and (r) incrementing by one the sequential position value of each said tail element, the incremented position value of each said tail element being the number of said data elements composing the homogeneous filament with which said tail element is associated.

Claim 13

The method of claim 6 wherein the value of each said first pointer identifies the processor in whose memory means the data element identified by said first pointer is stored.

Claim 14

The method of claim 6 wherein said data elements are aligned in the respective memory means of said processors such that information comprising each said data element is stored at the same address location.

Claim 15

The method of claim 7 including the additional step, following step (g), of: transferring the incremented position value respectively from each said tail element to the processor having a numerical address identity equal to the element value of the transferring tail element.

Claim 16

Apparatus for determining the number of times each of a plurality of different element values occurs in a group of data elements, each said data element including one of said element values, said apparatus comprising distributed processing means for: (a) forming said group of data elements into a data string in which each said data element includes a first pointer identifying an immediately preceding element closer to a true head element in said data string; (b) partitioning said data string into a plurality of homogeneous filaments each composed only of said data elements having the same element value, so that there is only one of said element values corresponding to each said homogeneous filament; and (c) determining the number of said data elements composing each said homogeneous filament, the number of said data elements so determined for each said homogeneous filament being the number of times the element value corresponding to that homogeneous filament occurs.

Claim 17

Apparatus for determining the number of times each of a plurality of different element values occurs in a group of data elements, each said data element including one of said element values, said apparatus comprising distributed processing means for: (a) forming said group of data elements into a data string in which each said data element includes a first pointer identifying an immediately preceding element closer to a true head element in said data string; (b) partitioning said data string into a plurality of homogeneous filaments each composed only of said data elements having the same element value, so that there is only one of said element values corresponding to each said homogeneous filament; and (c) determining the number of said data elements composing each said homogeneous filament, the number of said data elements so determined for each said homogeneous filament being the number of times the element value corresponding to that homogeneous filament occurs.

Claim 18

The apparatus of claim 14, said number of data element determining means of said distributed processing means further comprising means for: (d) identifying a tail element of each said homogeneous filament; (e) assigning a sequential position value to each said data element in each said homogeneous filament, said true head element of each said homogeneous filament being assigned the sequential position value of zero, the sequential position value of each said data element exceeding by one the sequential position value of the data element identified by its first pointer; and (f) incrementing by one the sequential position value of each said tail element, the incremented position value of each said tail element being the number of said data elements composing the homogeneous filament with which said tail element is associated.

Claim 19

The apparatus of claim 14, said number of data element determining means of said distributed processing means further comprising means for: (d) identifying a tail element of each said homogeneous filament; (e) copying said first pointer into a second pointer at each said data element; (f) defining at each said true head element a sequential position value of zero; (g) setting a tag at said true head element; (h) setting a counter value at zero; (i) determining if said tag is set at each said data element in each said homogeneous filament and, if so, proceeding to (o) and, if not, proceeding to (j); (j) taking a value 2 i +v as the sequential position value at each said data element for which said tag is not set, each of the last recited data elements being a receiving element, where: i=the current counter value; and v=the sequential position value of a passing element identified by said second pointer of said receiving element; (k) copying at each said receiving element the tag of said passing element; (l) determining, at each said data element for which said tag is not set, the data element currently identified by said second pointer of said determining element and substituting said second pointer of said currently identified data element for said second pointer of said determining element; (m) incrementing by one said counter value; (n) returning to (i); and (o) incrementing by one said sequential position value of each said tail element, the incremented position value of each said tail element being the number of said data elements composing the homogeneous filament with which said tail element is associated.

Claim 20

The apparatus of claim 14 wherein said element value for each said data element is expressed in binary form with the same bit length, a true head indicator being set at said true head element of said data string, said data string partitioning means of said distributed processing means further comprising means for: (d) specifying a predetermined one of the bits of said element value as a partitioning bit, said data string being an original string composed of at least one string fragment, each said fragment being composed of a maximal continuous sequence of said data elements having the same value of said partitioning bit; (e) copying at all said data elements the value of said first pointer into a second pointer; (f) determining for all said data elements except said true head element if the immediately preceding element has the same value of said partitioning bit as itself; (g) setting a first tag at all said data elements for which it is determined in (f) that the immediately preceding element does not have the same partitioning bit value as itself, each said data element having said first tag so set being a fragment head element; (h) setting said first tag and a second tag only at said true head element, each said data element having said first tag set or not set being respectively a tagged or untagged element; (i) getting, for all said untagged elements, the values of said first and second tags and said second pointer from the data element currently identified by said second pointer of said untagged element; (j) substituting, at all said untagged elements, the values of said first and second tags and said second pointer obtained in (i) from said currently identified element respectively for said first and second tags and said second pointer of said untagged element; (k) repeating (i) and (j) until all of said data elements in said original string are tagged; (l) setting said true head indicator at each said fragment head element for which it is determined that said second tag is set at the immediately preceding element in said original string; (m) getting, only for each said fragment head element, the value of said second pointer of said immediately preceding element in said original string and substituting the last recited second pointer value for the value of said first pointer of each said fragment head element, so that two filaments are formed each being defined by said first pointer at each said data element and said true head indicator being set at the data element at the head of each said filament; and (n) specifying as said partitioning bit a different one of the bits of said element value and repeating (e) through (m) unless all bits of said element value have been so specified, (e) through (m) being repeated with each said filament formed in the last performance of (m) being said original string, (b) being complete if all bits of said element value have been specified as said partitioning bit whereupon each said filament formed in the last performance of (m) is homogeneous.

Claim 21

The apparatus of claim 17, said number of data element determining means of said distributed processing means further comprising means for: (o) identifying a tail element of each said homogeneous filament; (p) assigning a sequential position value to each said data element in each said homogeneous filament, said true head element of each said homogeneous filament being assigned the sequential position value of zero, the sequential position value of each said data element exceeding by one the sequential position value of the data element identified by its first pointer; and (q) incrementing by one the sequential position value of each said tail element, the incremented position value of each said tail element being the number of said data elements composing the homogeneous filament with which said tail element is associated.

Claim 22

Apparatus for determining the number of times each of a plurality of different element values occurs in a group of data elements, each said data element including one of said element values, said apparatus comprising: means for forming said group of data elements into a data string in which each said data element includes a first pointer identifying an immediately preceding element closer to a true head element in said data string; means for partitioning said data string into a plurality of homogeneous filaments each composed only of said data elements having the same element value, so that there is only one of said element values corresponding to each said homogeneous filament; and means for determining the number of said data elements composing each said homogeneous filament, the number of said data elements so determined for each said homogeneous filament being the number of times the element value corresponding to that homogeneous filament occurs.

Claim 23

Apparatus for determining the number of times each of a plurality of different element values occurs in a group of data elements, each said data element including one of said element values, said apparatus comprising: means for forming said group of data elements into a data string in which each said data element includes a first pointer identifying an immediately preceding element closer to a true head element in said data string; means for partitioning said data string into a plurality of homogeneous filaments each composed only of said data elements having the same element value, so that there is only one of said element values corresponding to each said homogeneous filament; and means for determining the number of said data elements composing each said homogeneous filament, the number of said data elements so determined for each said homogeneous filament being the number of times the element value corresponding to that homogeneous filament occurs.

Claim 24

The apparatus of claim 19, said means for determining the number of said data elements further comprising: means for identifying a tail element of each said homogeneous filament; means for assigning a sequential position value to each said data element in each said homogeneous filament, said true head element of each said homogeneous filament being assigned the sequential position value of zero, the sequential position value of each said data element exceeding by one the sequential position value of the data element identified by its first pointer; and means for incrementing by one the sequential position value of each said tail element, the incremented position value of each said tail element being the number of said data elements composing the homogeneous filament with which said tail element is associated.

Claim 25

Apparatus for determining the number of times each of a plurality of different element values occurs in a group of data elements on a distributed processing system, each said data element including one of said element values, said processing system comprising a plurality of processors intercoupled to communicate with one another, each said processor comprising processing means for performing arithmetic and logic functions and memory means for storing information, said processing system being responsive to a single sequence of instructions applied in parallel to all of said processors, said distributed processing system being programmed to perform the steps of: (a) storing each said data element in the memory means of a different one of said processors; (b) forming said group of data elements into a data string in which each said data element includes a first pointer identifying an immediately preceding element closer to a true head element in said data string; (c) partitioning said data string into a plurality of homogeneous filaments each composed only of said data elements having the same element value, so that there is only one of said element values corresponding to each said homogeneous filament; and (d) determining the number of said data elements composing each said homogeneous filament, the number of said data elements so determined for each said homogeneous filament being the number of times the element value corresponding to that homogeneous filament occurs.

Claim 26

Apparatus for determining the number of times each of a plurality of different element values occurs in a group of data elements on a distributed processing system, each said data element including one of said element values, said processing system comprising a plurality of processors intercoupled to communicate with one another, each said processor comprising processing means for performing arithmetic and logic functions and memory means for storing information, said processing system being responsive to a single sequence of instructions applied in parallel to all of said processors, said distributed processing system being programmed to perform the steps of: (a) storing each said data element in the memory means of a different one of said processors; (b) forming said group of data elements into a data string in which each said data element includes a first pointer identifying an immediately preceding element closer to a true head element in said data string; (c) partitioning said data string into a plurality of homogeneous filaments each composed only of said data elements having the same element value, so that there is only one of said element values corresponding to each said homogeneous filament; and (d) determining the number of said data elements composing each said homogeneous filament, the number of said data elements so determined for each said homogeneous filament being the number of times the element value corresponding to that homogeneous filament occurs.

Claim 27

The apparatus of claim 21, said distributed processing system being further programmed to perform the substeps, of step (d), of: (e) identifying a tail element of each said homogeneous filament; (f) assigning a sequential position value to each said data element in each said homogeneous filament, said true head element of each said homogeneous filament being assigned the sequential position value of zero, the sequential position value of each said data element exceeding by one the sequential position value of the data element identified by its first pointer; and (g) incrementing by one the sequential position value of each said tail element, the incremented position value of each said tail element being the number of said data elements composing the homogeneous filament with which said tail element is associated.

Claim 28

The apparatus of claim 21, said distributed processing system being further programmed to perform the substeps, of step (d), of: (e) identifying a tail element of each said homogeneous filament; (f) copying said first pointer into a second pointer at each said data element; (g) defining at each said true head element a sequential position value of zero; (h) setting a tag at said true head element; (i) setting a counter value at zero; (j) determining if said tag is set at each said data element in each said homogeneous filament and, if so, proceeding to step (p) and, if not, proceeding to step (k); (k) taking a value 2 i +v as the sequential position value at each said data element for which said tag is not set, each of the last recited data elements being a receiving element, where: i=the current counter value; and v=the sequential position value of a passing element identified by said second pointer of said receiving element; (l) copying at each said receiving element the passing element; (m) determining, at each said data element for which said tag is not set, the data element currently identified by said second pointer of said determining element and substituting said second pointer of said currently identified data element for said second pointer of said determining element; (n) incrementing by one said counter value; (o) returning to step (j); and (p) incrementing by one said sequential position value of each said tail element, the incremented position value of each said tail element being the number of said data elements composing the homogeneous filament with which said tail element is associated.

Claim 29

The apparatus of claim 21 wherein said element value for each said data element is expressed in binary form with the same bit length, a true head indicator being set at said true head element of said data string, said distributed processing system being further programmed to perform the substeps, of step (c), of: (e) specifying a predetermined one of the bits of said element value as a partitioning bit, said data string being an original string composed of at least one string fragment, each said fragment being composed of a maximal continuous sequence of said data elements having the same value of said partitioning bit; (f) copying at all said data elements the value of said first pointer into a second pointer; (g) determining for all said data elements except said true head element if the immediately preceding element has the same value of said partitioning bit as itself; (h) setting a first tag at all said data elements for which it is determined in step (g) that the immediately preceding element does not have the same partitioning bit value as itself, each said data element having said first tag so set being a fragment head element; (i) setting said first tag and a second tag only at said true head element, each said data element having said first tag set or not set being respectively a tagged or untagged element; (j) getting, for all said untagged elements, the values of said first and second tags and said second pointer from the data element currently identified by said second pointer of said untagged element; (k) substituting, at all said untagged elements, the values of said first and second tags and said second pointer obtained in step (j) from said currently identified element respectively for said first and second tags and said second pointer of said untagged element; (l) repeating steps (j) and (k) until all of said data elements in said original string are tagged; (m) setting said true head indicator at each said fragment head element for which it is determined that said second tag is set at the immediately preceding element in said original string; (n) getting, only for each said fragment head element, the value of said second pointer of said immediately preceding element in said original string and substituting the last recited second pointer value for the value of said first pointer of each said fragment head element, so that two filaments are formed each being defined by said first pointer at each said data element and said true head indicator being set at the data element at the head of each said filament; and (o) specifying as said partitioning bit a different one of the bits of said element value and repeating steps (f) through (n) unless all bits of said element value have been so specified, steps (f) through (n) being repeated with each said filament formed in the last performance of step (n) being said original string, step (c) being complete if all bits of said element value have been specified as said partitioning bit whereupon each said filament formed in the last performance of step (n) is homogeneous.

Claim 30

The apparatus of claim 24, said distributed processing system being programmed to perform the substeps, of step (d), of: (p) identifying a tail element of each said homogeneous filament; (q) assigning a sequential position value to each said data element in each said homogeneous filament, said true head element of each said homogeneous filament being assigned the sequential position value of zero, the sequential position value of each said data element exceeding by one the sequential position value of the data element identified by its first pointer; and (r) incrementing by one the sequential position value of each said tail element, the incremented position value of each said tail element being the number of said data elements composing the homogeneous filament with which said tail element is associated.

Claim 31

The apparatus of claim 22, said distributed processing system being programmed to perform the additional step, following step (g), of: transferring the incremented position value respectively from each said tail element to the processor having a numerical address identity equal to the element value of the transferring tail element.