System for texture-based automatic detection of man-made objects in representations of sensed natural environmental scenes

• US 5,341,439 A
• Filed: 08/21/1992
• Issued: 08/23/1994
• Est. Priority Date: 09/21/1989
• Status: Expired due to Term

• Current Assignee: Shin-Yi Hsu
• Original Assignee: Shin-Yi Hsu
• Inventors: Hsu, Shin-Yi
• Primary Examiner(s): Mancuso, Joseph
• Assistant Examiner(s): Fox, David
• Application Number:
  US07/934,094
• Time in Patent Office: 732 Days
• Field of Search: 382/1, 382/19, 382/28, 382/27, 382/49, 358/107, 358/108, 358/109, 395/130, 348/135, 348/143, 348/144
• US Class Current: 382/103
• CPC Class Codes: G06T 2207/20021 Dividing image into blocks, subimages or windows
• CPC Class Codes: G06T 2207/20104 Interactive definition of region of interest [ROI]
• CPC Class Codes: G06T 7/41 based on statistical description of texture
• CPC Class Codes: G06V 10/255 Detecting or recognising potential candidate objects based on visual cues, e.g. shapes

Abstract

A system for texture-based automatic detection of man-made objects or targets or features of scenes (automatic target recognition, ATR or detection, ATD) in representations of sensed natural environmental scenes spatially partitions digitized picture elements (pixels) of the scene into plural spatially coordinated groups of pixels and simultaneously determines texture measures including a composite texture measure for each group of pixels. Following self-calibration of the spatially coordinated composite texture measure values along spatially delineated row and orthogonal column directions of groups of pixels, areas of interest (AOIs) are identified as the groups of pixels most likely to contain a man-made object, target or feature in a decision logic which performs a group of statistical tests. Automatic detection of the spatial location of a man-made object, target, or feature within each AOI occurs by a single-threshold segmentation of pixels associated with each AOI into a grouping of target pixels and a grouping of non-target pixels. The spatial location of the man-made object, target, or feature within each AOI can be visually displayed, and the target location coordinates within the entire scene reported, whereupon the automatic detection system is reset for sensing and processing of a subsequent scene.

Claims

1. A method for texture-based automatic detection of targets comprising the steps of:

sensing a natural environment scene;

digitizing the sensed scene into a first plurality of spatially coordinated picture elements (pixels) with associated digital signals;

storing the spatially coordinated digital signals;

partitioning the first plurality of spatially coordinated pixels into a second plurality of spatially overlapping groups of pixels arranged spatially sequentially in orthogonal rows and columns with an identical fraction of pixels of the first plurality of pixels in each group of pixels, said spatially sequentially arranged groups having a sequential arrangement extending along successive straight paths across the scene;

storing the digital signals associated with the partitioned groups of pixels;

determining simultaneously a plurality of texture measure values and a composite texture measure value from the digital signals associated with each partitioned spatially coordinated group of pixels;

storing the spatially coordinated composite texture measure values of the groups of pixels; and

identifying said target in the scene from said texture measure values, and wherein said identifying step includes the steps of;

self calibrating the composite texture measure values of the stored spatially coordinated groups of pixels along both the rows and the columns;

storing the spatially coordinated self-calibrated composite texture measure values;

deciding the identity of an area or areas of interest (AOI) as the spatially coordinated group or groups of pixels most likely to contain a target;

routing the AOI identity decision obtained in said deciding step;

detecting automatically the presence of a target within the AOI or AOIs;

displaying the spatial location coordinates of the target within the AOI or AOIs;

reporting the spatial location coordinates of the target or targets; and

resetting the automatic detection system for acceptance of a subsequent representation of a sensed scene.

2. The method according to claim 1, wherein said partitioning step comprises:

selecting at least ten of the second plurality of spatially overlapping groups of pixels arranged spatially sequentially across the scene with each group of pixels having an identical fraction of pixels of the first plurality of spatially coordinated pixels.

3. The method according to claim 1, wherein said determining step comprises:

establishing simultaneously for each group of pixels at a time a third plurality of numbers n_a, n_b, n_c, through n_i+1 of segmented sub-regions of pixels at each of a fourth plurality of temporally simultaneous threshold levels with incremental increases V_i, where i is a plurality of successive integers, for the digital signals associated with the pixels in each group;

forming a fifth plurality of texture measure values R₁, R₂, through R_i simultaneously for each group of pixels at a time in accordance with the following relationships, ##EQU3## generating a composite texture measure value G as the linear sum of texture measure values R₁, R₂, through R_i for each group of pixels at a time.

4. The method according to claim 1, wherein said self-calibrating step comprises:

designating the composite texture measure value G_1r of a first group in the spatial sequence of groups of pixels in each row r as a ratio G_1r /G_1r =1.00;

deriving self-calibrated composite texture measure values for each successive neighboring group of pixels in each row r of groups by the ratio of composite texture measure values between the successive neighboring group and the preceding group in that row;

designating the composite texture measure value G_1c of a first group in the spatial sequence of groups of pixels in each column c as a ratio G_1c /G_1c =1.00; and

deriving self-calibrated composite texture measure values for each successive neighboring group of pixels in each column c of groups of pixels by the ratio of composite texture measure values between the successive neighboring group and the preceding group in that column.

5. The method according to claim 1, wherein said deciding step comprises:

subjecting the row and column spatially coordinated self-calibrated composite texture measure values to a group of statistical tests to identify an AOI or AOIs.

6. The method according to claim 5, wherein said detecting step comprises:

retrieving digital signals associated with partitioned AOI groups of pixels stored in said partitioned pixel group storing step;

choosing a single threshold level; and

separating the spatially coordinated AOI digital signals retrieved in said retrieving step into a first group of AOI pixel signals at or below said single threshold level and into a second group of substantially target pixel signals above said single threshold level.

7. The method according to claim 6, wherein said displaying step comprises:

rendering the spatial extent of the target within the AOI as a clustered arrangement of different visually perceived colors indicative of different ranges of target pixel signals in the second group of pixels separated in said separating step; and

showing the spatial extent of the first group of pixels separated in said separating step in yet another perceived color.

8. The method according to claim 2, wherein said selecting step further comprises:

arranging each of said second plurality of spatially coordinated groups of pixels in a format; and

deploying an overlap of between one percent and fifty percent of the number of pixels spatially associated with the format arranged in said arranging step among neighboring groups of pixels of said second plurality of spatially sequentially arranged coordinated groups of pixels.

9. The method according to claim 3, wherein each of said establishing and forming and generating steps further comprises:

determining said texture measure values and composite texture measure values within the incrementally increasing range of threshold levels V_i, where i≦

20.

10. The method according to claim 3, wherein said establishing step further comprises:

setting the incremental increase of threshold levels from V_i as a linear incremental increase.

11. The method according to claim 4, wherein each of said deriving steps further comprises:

electing second-neighbor groups of pixels in the rows and columns of groups of pixels for derivation of the self-calibrated composite texture measure values.

12. The method according to claim 5, wherein said subjecting step further comprises:

applying tests selected from the group consisting of a rank order test of self-calibrated composite texture measure values;

a linear correlation test between ranked values and neighboring unranked values in row and column directions; and

a sum-of-squares slope difference test between ranked values and unranked values in row and column directions.

13. A system for texture-based automatic detection of targets comprising:

means for sensing a natural environment scene;

means for digitizing the sensed scene into a first plurality of spatially coordinated picture elements (pixels) with associated digital signals;

means for storing the spatially coordinated digital signals;

means for partitioning the first plurality of spatially coordinated pixels into a second plurality of overlapping groups of pixels arranged spatially sequentially in rows and orthogonal columns with an identical fraction of pixels of the first plurality of pixels in each group of pixels, said spatially sequentially arranged groups having a sequential arrangement extending along successive straight paths across the scene;

means for storing the digital signals associated with the partitioned groups of pixels;

means for determining simultaneously a plurality of texture measure values and a composite texture measure value from the digital signals associated with each partitioned spatially coordinated group of pixels;

means for storing the spatially coordinated composite texture measure values of the groups of pixels; and

means for identifying said target in the scene from said texture measure values, and further comprising;

means for self calibrating the composite texture measure values of the stored spatially coordinated groups of pixels along both the rows and the columns;

means for storing the spatially coordinated self-calibrated composite texture measure values;

means for deciding the identity of an area or areas of interest (AOI) as the spatially coordinated group or groups of pixels most likely to contain a target;

means for routing the AOI identity decision obtained by said deciding means;

means for detecting automatically the presence of a target within the AOI or AOIs;

means for displaying the spatial location coordinates of the target within the AOI or AOIs;

means for reporting the spatial location coordinates of the target or targets; and

means for resetting the automatic detection system for acceptance of a subsequent representation of a sensed scene.

14. The system according to claim 13, wherein said partitioning means comprises:

means for selecting at least ten of the second plurality of spatially overlapping groups of pixels arranged spatially sequentially across the scene with each group of pixels having an identical fraction of pixels of the first plurality of spatially coordinated pixels.

15. The system according to claim 13, wherein said determining means comprises:

means for establishing simultaneously for each group of pixels at a time a third plurality of numbers n_a, n_b, n_c, through n_i+1 of segmented sub-regions of pixels at each of a fourth plurality of temporally simultaneous threshold levels with incremental increase V_i, where i are successive integers, for the digital signals associated with the pixels in each group;

means for forming a fifth plurality of texture measure values R₁, R₂, through R_i simultaneously for each group of pixels at a time in accordance with the following relationships, ##EQU4## means for generating a composite texture measure value G as the linear sum of texture measure values R₁, R₂, through R_i for each group of pixels at a time.

16. The system according to claim 13, wherein said self-calibrating means comprises:

means for designating the composite texture measure value G_1r of a first group in the spatial sequence of groups of pixels in each row r as a ratio G_1r /G_1r =1.00;

means for deriving self-calibrated composite texture measure values for each successive neighboring group of pixels in each row r of groups of pixels by the ratio of composite texture measure values between the successive neighboring group and the preceding group in that row;

means for designating the composite texture measure value G_1c of a first group in the spatial sequence of groups of pixels in each column c as a ratio G_1c /G_1c =1.00; and

means for deriving self-calibrated composite texture measure values for each successive neighboring group of pixels in each column c of groups of pixels by the ratio of composite texture measure values between the successive neighboring group and the preceding group in that column.

17. The system according to claim 13, wherein said deciding means comprises:

means for subjecting the row and column spatially coordinated self-calibrated composite texture measure values to a group of statistical tests to identify an AOI or AOIs.

18. The system according to claim 17, wherein said detecting means comprises:

means for retrieving digital signals associated with partitioned AOI groups of pixels stored in said partitioned pixel storage means;

means for choosing a single threshold level; and

means for separating the spatially coordinated AOI digital signals retrieved by said retrieving means into a first group of AOI pixel signals at or below said single threshold level and into a second group of substantially target pixel signals above said single threshold level.

19. The system according to claim 18, wherein said displaying means comprises:

means for rendering the spatial extent of the target within the AOI as a clustered arrangement of different visually perceived colors indicative of different ranges of target pixel signals in the second group of pixels separated by said separating means; and

means for showing the spatial extent of the first group of pixels separated by that separating means in yet another perceived color.

20. The system according to claim 15, wherein said:

means for establishing includes means for establishing up to twenty temporally simultaneous threshold levels which incrementally increase by one.

21. The system according to claim 16, wherein said

means for deriving includes means for deriving the row and column self-calibrated composite texture measure values by the ratio between each second-neighbor successive group and each second-neighbor preceding group.

22. The system according to claim 13, wherein said storing means comprises:

means for electronic storing of digital data; and

means for resetting said electronic storing means; and

said reporting and resetting means further comprises;

means for remote non-verbal communication.

23. The system according to claim 17, wherein said subjecting means comprises:

means for applying tests selected from the group consisting of a rank order test of self-calibrated composite texture measures;

a linear correlation test between ranked values and neighboring unranked values in row and column directions; and

a sum-of-squares slope difference test between ranked values and unranked values in row and column directions.