1. Archaeological Land Use Characterization using Multispectral Remote Sensing Data Dr. Iván Esteban Villalón Turrubiates, Member, IEEE María de Jesús Llovera Torres UNIVERSIDAD DE GUADALAJARA CENTRO UNIVERSITARIO DE LOS VALLES Monitoring Hidrological Variations using Multispectral SPOT-5 Data: Regional Case of Jalisco in Mexico Dr. Iván Esteban Villalón Turrubiates, Member, IEEE

2. Overview - Abstract - Remote Sensing Definition - Sensor Resolution - Introduction to Image Classification - Model Formalism - Verification Protocols - Simulation Experiments - Concluding Remarks

3. Abstract  Proposition - A new and efficient classification approach of remote sensing signatures extracted from large-scale multispectral imagery.  Contribution - This approach exploits the idea of combining the spectral signatures from a remote sensing image to perform a novel and accurate classification technique.  Verification - Simulation results are provided to verify the efficiency of the proposed approach.

4. REMOTE SENSING DEFINITION

5. Remote Sensing  Remote Sensing can be defined as:  "The arte and science to obtain data from an object avoiding direct contact with it” (Jensen 2000).  There is a transmission medium involved?

7. Remote Sensing  Of the Environment:  … is the collection of information regarding our Planet surface and its phenomena involving sensors that are not in direct contact with the studied area. The main focus is in recollected information from a spatial perspective throughout electromagnetic radiation transmission.

9. Remote Sensing  Sensor election.  Reception, storage and digital signal processing of the data.  Analysis of the resulting information.

10. A) Illumination Source B) Radiation C) Interaction with the object D) Radiation sensing E) Transmission, reception and data processing F) Analysis and interpretation G) Application Process

11. SENSOR RESOLUTION

12. Resolution  All remote sensing systems use four types of resolution:  Spatial  Spectral  Temporal  Radiometric

13. Spatial Resolution

14. Spectral Resolution

15. Time July 1 July 12 July 23August 3 11 days 16 days July 2July 8August 3 Temporal Resolution

16. 6-bits Range 0 63 8-bits Range 0 255 0 10-bits Range 1023 Radiometric Resolution

17. INTRODUCTION TO IMAGE CLASSIFICATION

18. Image Classification  Why classify?  Make sense of a landscape  Place landscape into categories (classes)  Forest, Agriculture, Water, Soil, etc.  Classification scheme = structure of classes  Depends on needs of users.

19. Typical uses  Provide context  Landscape planning or assessment  Research projects  Natural resources management  Archaeological studies  Drive models  Meteorology  Biodiversity  Water distribution  Land use

20. Example: Near Mary’s Peak Derived from a 1988 Landsat TM image Distinguish types of forest Open Semi-open Broadleaf Mixed Young Conifer Mature Conifer Old Conifer Legend

21. Classification: Critical Point  LAND COVER not necessarily equivalent to LAND USE  We focus on what’s there: LAND COVER  Many users are interested in how what is there is being used: LAND USE  Example  Grass is land cover; pasture and recreational parks are land uses of grass

22. Basic Strategy: How to do it?  Use radiometric properties of remote sensor  Different objects have different spectral signatures

23.  In an easy world, all “vegetation” pixels would have exactly the same spectral signature.  Then we could just say that any pixel in an image with that signature was vegetation.  We could do the same for soil, water, etc. to end up with a map of classes. Basic Strategy: How to do it?

24. But in reality, that is not the case. Looking at several pixels with vegetation, you’d see variety in spectral signatures. The same would happen for other types of pixels, as well. Basic Strategy: How to do it?

25. The Classification Trick: Deal with variability Different ways of dealing with the variability lead to different ways of classifying images. To talk about this, we need to look at spectral signatures a little differently.

26. Think of a pixel’s brightness in a 2-Dimensional space. The pixel occupies a point in that space. The vegetation pixel and the soil pixel occupy different points in a 2-D space.

27. With variability, the vegetation pixels now occupy a region, not a point, of n-Dimensional space. Soil pixels occupy a different region of n-Dimensional space.

28. Classification: Delineate boundaries of classes in n-dimensional space Assign class names to pixels using those boundaries Basic Strategy: Deal with variability

29. Classification Strategies  Two basic strategies:  Supervised Classification  We impose our perceptions on the spectral data.  Unsupervised Classification  Spectral data imposes constraints on our interpretation.

30. Digital Image Supervised Classification The computer then creates... Supervised classification requires the analyst to select training areas where he knows what is on the ground and then digitize a polygon within that area… Mean Spectral Signatures Known Conifer Area Known Water Area Known Deciduous Area Conifer Deciduous Water

31. Multispectral Image Information (Classified Image) Mean Spectral Signatures Spectral Signature of Next Pixel to be Classified Conifer Deciduous Water Unknown Supervised Classification

32. Water Conifer Deciduous Legend: Land Cover Map The Result: Image Signatures

33. Unsupervised Classification  In unsupervised classification, the spectral data imposes constraints on our interpretation.  How? Rather than defining training sets and carving out pieces of n-Dimensional space, we define no classes beforehand and instead use statistical approaches to divide the n-Dimensional space into clusters with the best separation.  After the fact, we assign class names to those clusters.

34. Unsupervised Classification Digital Image The analyst requests the computer to examine the image and extract a number of spectrally distinct clusters… Spectrally Distinct Clusters Cluster 3 Cluster 5 Cluster 1 Cluster 6 Cluster 2 Cluster 4

35. Saved Clusters Cluster 3 Cluster 5 Cluster 1 Cluster 6 Cluster 2 Cluster 4 Unsupervised Classification Output Classified Image Unknown Next Pixel to be Classified

36. Unsupervised Classification Conif. Hardw. Water Land Cover Map Legend Water Conifer Hardwood Labels It is a simple process to regroup (recode) the clusters into meaningful information classes (the legend). The result is essentially the same as that of the supervised classification:

37. MODEL FORMALISM

38. Multispectral Imaging  Is a technology originally developed for space-based imaging.  Multispectral images are the main type of images acquired by remote sensing radiometers.  Usually, remote sensing systems have from 3 to 7 radiometers; each one acquires one digital image in a small band of visible spectra, ranging 450 to 690 nm, called red-green-blue (RGB) regions:  Blue -> 450-520 nm.  Green -> 520-600 nm.  Red -> 600-690 nm.  The combination of the RGB spectral bands generates the so-called True-Color RS images.

39.  Statistical Approach.  Assume normal distributions of pixels within classes.  For each class, build a discriminant function  For each pixel in the image, this function calculates the probability that the pixel is a member of that class.  Takes into account mean and variance of training set.  Each pixel is assigned to the class for which it has the highest probability of membership. Weighted Pixel Statistics Method

40. BlueGreenRedNear-IRMid-IR Mean Signature 1 Candidate Pixel Mean Signature 2 It appears that the candidate pixel is closest to Signature 1. However, when we consider the variance around the signatures… Relative Reflectance Weighted Pixel Statistics Method

41. BlueGreenRedNear-IRMid-IR Mean Signature 1 Candidate Pixel Mean Signature 2 The candidate pixel clearly belongs to the signature 2 group. Relative Reflectance Weighted Pixel Statistics Method

44. VERIFICATION PROTOCOLS

45. Verification Protocols  A set of three synthesized images are used as verification protocols.  All synthesized images are True-Color (RGB), presented in 1024-by-1024 pixels (TIFF format).  Each synthesized image contains three different regions (in yellow, blue and black colors) with a different pattern.  The developed Weighted Pixel Statistics (WPS) algorithm is compared with the most traditional Weighted Order Statistics (WOS) method [S.W. Perry, H.S. Wong, 2002].

46. Results: 1 st Synthesized Scene Synthesized Scene WOS Classification WPS Classification

47. Quantitative Comparison 1 st Synthesized Scene

48. Results: 2 nd Synthesized Scene Synthesized Scene WOS Classification WPS Classification

49. Qualitative Comparison 2 nd Synthesized Scene Synthesized Scene WOS Classification WPS Classification

50. Quantitative Comparison 2 nd Synthesized Scene

51. Results: 3 rd Synthesized Scene Synthesized Scene WOS Classification WPS Classification

52. Qualitative Comparison 3 rd Synthesized Scene Synthesized Scene WOS Classification WPS Classification

53. Quantitative Comparison 3 rd Synthesized Scene

54. Remarks  The quantitative study is performed calculating the classified percentage obtained with the WOS and WPS methods, respectively.  The WOS method uses only 1 spectral band.  The WPS method uses the information from the three spectral bands to analyze the pixel-level neighborhood means and variances.  The results shows a more accurate and less smoothed identification of the classes.

55. SIMULATION EXPERIMENTS

56. Archaeological Land Use  A Remote Sensing Signatures (RSS) electronic map is extracted from the multispectral image. Three level RSS are selected for this particular simulation process, defined as:  ██ – Archaeological land use zones.  ██ – Modern land use zones.  ██ – Natural land cover zones.  ██ – Unclassified zones.

57. Archaeological Site "Guachimontones", Jalisco Mexico

58. Simulation Results Scene from "Guachimontones" Original Scene WPS Classification

59. Hidrological Variations  A Remote Sensing Signatures (RSS) electronic map is extracted from the multispectral image. Three level RSS are selected for this particular simulation process, defined as:  ██ – Humid zones.  ██ – Dry zones.  ██ – Wet zones.  ██ – Unclassified zones.

60. Simulation Results Scene from "La Vega" dam, Jalisco Mexico Original Scene WPS Classification

61. CONCLUDING REMARKS

62. Remarks  The WOS classifier generates several unclassified zones because it uses only one spectral band in the classification process.  The WPS classifier provides a high-accurate classification without unclassified zones because it uses more robust information in the processing.  The qualitative and quantitative analysis probe the efficiency of the proposed approach.

63. Future Work  Comparison with several classification techniques.  A more extensive performance analysis of the proposed approach with different synthesized images.  Application to remote sensing imagery and the study of its performance.  Hardware implementation of the proposed approach.

64. Dr. Iván Esteban Villalón Turrubiates, Member, IEEE UNIVERSIDAD DE GUADALAJARA CENTRO UNIVERSITARIO DE LOS VALLES THANK YOU! Questions?