1. 1 Prepared by: Precise Object Tracking under Deformation Eng. Mohamed Hassan, EAEA Supervised by: Prof. Dr. Hussien Konber, Al Azhar University Prof. Dr. Mohamoud Ashour, EAEA Dr. Ashraf Aboshosha, EAEA Submitted to: Communication & Electronics Dept., Al Azhar University

2. 2 Key subjects of this framework include:  Motivation  Visual tracking applications  Block diagram of object tracking system  Image deformation types  Object extraction  Morphological operations  Geometrical Modeling and pose estimation  Conclusion and Future Work Outlines

3. 3 Motivation The main objectives of this research work are to:  Overcome the imprecision in object tracking caused by different deformation sources such as noise, change of illumination, blurring, scaling and rotation.  Developing a three dimensional (3D) geometrical model to determine the current pose of an object and predict its future location based on FIR model  Presenting a robust ranging technique to track a visual target instead of the traditional expensive ranging sensors.

4. 4  The precise object tracking is an essential issue in several applications such as:  Robot vision  Automated surveillance (civil and military)  Medical applications  Satellite and space systems  Traffic systems  Security etc. Visual Tracking Applications

5. 5 Block Diagram of Object Tracking System Video Camera USB Camera USB Bus Frame grabber PC Image Acquisition Image Processing Output Target

6. 6 Image Deformation Types  Noise.  Scaling &Rotation.  Blurring  Change of illumination.

7. 7 Definition: is considered to be any measurement that is not part of the phenomena of interest. Images are affected by different types of noise:  Gaussian noise  Salt and Pepper noise  Poisson Noise  Speckle Noise Image Deformation: Noise

8. 8 Image De-noising Techniques The following digital filters have been employed for denoising  Linear filter (Average filter, Gaussian filter and unsharp filter)  Non linear filter (Median filter and Adaptive filter)  Coiflet Wavelets  Proposed filter

9. 9  Spatial filtering term is the filtering operations that are performed directly on the pixels of an image.  The process consists simply of moving the filter mask from point to point in an image.  At each point (x,y) the response of the filter at that point is calculated using a predefined relationship. Spatial Filters

10. 10 f(x-1,y-1)f(x-1,y)f(x-1,y+1) f(x,y-1)f(x,y)f(x,y+1) f(x+1,y-1)f(x+1,y)f(x+1,y+1) w(-1,-1)w(-1,0)w(-1,1) w(0,-1)w(0,0)w(0,1) w(1,-1)w(1,0)w(1,1) The result is the sum of products of the mask coefficients with the corresponding pixels directly under the mask Pixels of image Mask coefficients w(-1,-1)w(-1,0)w(-1,1) w(0,-1)w(0,0)w(0,1) w(1,-1)w(1,0)w(1,1) Linear Spatial Filters

11. 11  Nonlinear spatial filters also operate on neighborhoods, and the mechanics of sliding a mask past an image are the same as was just outlined.  The filtering operation is based conditionally on the values of the pixels in the neighborhood under consideration.  Order-statistics filters are nonlinear spatial filters whose response is based on ordering (ranking) Nonlinear Spatial Filters

12. 12  The Wavelet transform is a multiresolution analysis tool which decomposes a signal into different frequency sub bands.  Wavelet transform, due to its excellent localization, has rapidly become an indispensable signal and image processing tool for a variety of applications.  Wavelet denoising attempts to remove the noise present in the signal while preserving the signal characteristics, regardless of its frequency content. Wavelet Transform

13. 13 Figure 1 The two-dimensional FWT - the analysis filter Wavelet Transform Figure 2 Two-scale of two-dimensional decomposition

14. 14 The proposed filter is a cascaded spatial filter based on median fitter and Coiflet wavelets. Its edge-preserving nature makes it useful in cases where edge blurring is undesirable. It is very useful in real object tracking. This filter is the best one for removing all types of noise Denoising Proposed Filter I/p image Median filter Coiflet WaveletsO/p image Figure 3 Cascaded spatial filter based on median fitter and Coiflet wavelets

15. 15 Image Similarity Measure To validate the efficiency of the previous digital filters the following similarity measures have been applied  2D Cross Correlation  Peak Signal-to-Noise Ratio (PSNR) dB

16. 16 2D Cross Correlation Unsharp filter Average filter Gaussian filter Median filter Adaptive filter Proposed filter Salt and paper noise 0.9234 0.98900.69830.98090.78040.9984 Gaussian noise 0.5651 0.98610.94460.9701 0.9876 Poisson noise 0.8270 0.99200.99000.99100.99130.9961 Speckle noise 0.6349 0.98790.7737 0.83410.85470.9871 Table 1. 2D cross correlation similarity measure

17. 17 Peak Signal-to-Noise Ratio (PSNR) dB Unsharp filter Average filter Gaussian filter Median filter Adaptiv e filter Proposed filter Salt and paper noise 18.59 27.37 25.4936.00 22.9749.48 Gaussian noise 9.9426.16 23.80 26.4226.79 32.80 Poisson noise 14.7428.71 30.21 31.9232.8043.16 Speckle noise 10.8626.7325.38 26.7127.59 37.67 Table 2. PSNR similarity measure

18. 18 Scaling & Rotation Definition: Scaling & rotation is affine Transformation where Straight lines remain straight, and parallel lines remain parallel. Scaling and Rotation: The linear transformation and radon transformation have been used to recover an image from a rotated and scaled origin.

19. 19 Scaled image Original image Scaled &rotated image Figure 4 Rotated and scaled image Scaling & Rotation

20. 20 Figure 5 Control point selection Linear Transformation

21. 21 Original image Scaled & rotated image recovered image Figure 6 Recovered by using linear transformation Linear Transformation

22. 22 Radon transform: This transform is able to transform two dimensional images with lines into a domain of possible line parameters, where each line in the image will give a peak positioned at the corresponding line parameters. Projections can be computed along any angle θ, by use general equation of the Radon transformation: Radon Transformation x' is the perpendicular distance of the beam from the origin and θ is the angle of incidence of the beams.

23. 23 Original image Figure7 Canny edge detection and edge linking Edge detectionEdge linking Radon Transformation

24. 24 Figure 8 Radon transform projections along 180 degrees, from -90 to +89 Radon Transformation

25. 25 Original image Rotated image recovered image Figure 9 Recovered by using radon transform Radon Transformation

26. 26 Blurring: degradation of an image can be caused by motion There are two types of blurring  Known blurring: the length and the angle of blurring are known  Unknown blurring: the length and the angle of blurring are unknown Blurring

27. 27 Deblurring Techniques  Deblurring using Wiener filter  Deblurring using a regularized filter  Deblurring using Lucy-Richardson algorithm  Deblurring using blind deconvolution algorithm A blurred or degraded image can be approximately described by this equation

28. 28 Deblurring using the Blind Deconvolution Algorithm Figure 10 Deblurring using the blind deconvolution algorithm

29. 29 Figure 11, Capability of object tracking under blurring (a, b) with known blur function and after deblurring (c, d (a) Blurred image (b) Person detection under motion deformation (c)Deblurred image (d) Person detection in deblurred image Deblurring Techniques

30. 30 Blurred image correlation with original one Deblurred image using correct parameters correlation Deblurring Techniques

31. 31 Deblurred image using longer PSF correlation Deblurred image using different angle correlation Figure 12, 2D cross correlation with the deblurring form Deblurring Techniques

32. 32 Correlation Condition blurred image with the original one 0.0614 deblurred image with the original one using correct parameters 0.3523 deblurred image with the original one using longer PSF 0.0558 deblurred image with the original one using different angle 0.1231 Table 3, 2D cross correlation with the deblurring form Deblurring Techniques

33. 33 Change of Illumination Change of illumination Color model deformation may happen due to the change in illumination Proposed solution Selecting an appropriate color model (RGB, HSV or yc b c r ) to overcome the deformation problem

34. 34 RGB Representation The RGB color model mapped to a cube A Representation of additive color mixing  Weak points of the RGB color model  RGB color model is affected by the change of illumination  RGB is non uniform color model

35. 35 HSV Representations  Hue, saturation and intensity are often plotted in cylindrical coordinates with hue the angle, saturation the radius, and intensity the axis. HSV color wheelconical representation of the HSV The cylindrical representation of the HSV

36. 36 Chrominance is defined as the difference between a color and a reference white at the same luminance. YC b C r Color Model The conversion from RGB to YC b C r The conversion from YC b C r to RGB

37. 37 Advantage of YC b C r The main advantages of this model are: The luminance component (Y) of YC b C r is independent of the color The skin color cluster is more compact in YC b C r than in other color space YC b C r has the smallest overlap between skin and non-skin data in under various illumination conditions. YC b C r is broadly utilized in video compression standards YC b C r is a family of color spaces used in video systems. YC b C r is one of two primary color spaces used to represent digital component video (the other is RGB).

38. 38  To track a visual target we have to relay on a segmentation technique such as:  Thresholding  Clustering  Region growing  Edge-based  Physical model-based  Frame Subtraction  Fast block matching Throughout this framework a color table thresholding segmentation technique has been applied to extract the visual target Object Extraction

39. 39 Original image sample Tracked object Homogeneous Object Extraction

40. 40 sample Homogeneous Object Extraction RGBYC b C r HSV Figure 13, Comparison of homogeneous object extraction Original image

41. 41 Original image Tracked object sample Inhomogeneous Object Extraction

42. 42 Original image RGB sample YC b C r HSV Figure 14, Comparison of inhomogeneous object extraction Inhomogeneous Object Extraction

43. 43 The most basic morphological operations are dilation and erosion Morphological operations  Dilation adds pixels to the boundaries of objects in an image.  Expand/enlarge objects in the image  Fill gaps or bays of insufficient width  Fill small holes of sufficiently small size  Connects objects separated by a distance less than the size of the window  Erosion removes pixels on object boundaries.  to erode away the boundaries of regions of foreground pixels (i.e. white pixels, typically).  Thus areas of foreground pixels shrink in size, and holes within those areas become larger

44. 44  Opening and Closing are morphological operations which are based on dilation and erosion.  Opening smoothes the contours of objects, breaks narrow isthmuses and eliminates thin protrusions.  Closing also produces the smoothing of sections of contours but fuses narrow breaks, fills gaps in the contour and eliminates small holes.  Opening is basically erosion followed by dilation while closing is dilation followed by erosion. Morphological operations

45. 45 Binary object Binary after removing extra pixel Binary object after dilation holes Binary object after closing Morphological operations Figure 15, The effect of the morphological operation

46. 46 Morphological operations Figure 16, Center of gravity, ellipse fitting and bound box of an image

47. 47 Geometrical Modeling Figure 17 object tracking at different distance

48. 48 Where, a = 30606.621 b=-0.03410108  The relation between distance (D) and no of pixel (N) Geometrical Modeling Figure 18. The relation between range (D) and projection size (N)

49. 49  The relation between the range and location of the object in 3D domain Geometrical Modeling Figure 19. The relation between the range and location of the object in 3D domain

50. 50 Motion Estimation and Prediction based on FIR Figure 19, FIR model structures

51. 51 Motion Estimation and Prediction based on FIR Figure 20, Models output w.r.t system output

52. 52 Motion Estimation and Prediction based on FIR Figure 21 Model output w.r.t system output

53. 53 Motion Estimation and Prediction based on FIR Figure 22 The capability of the model to predict the output if the system input is known

54. 54 Throughout this framework the following academic tasks have been achieved Developing a novel Universal filter for image denoising Selecting qualitative radon transformation for correction of the rotation Intensive comparative study for dealing with kwon/unknown bulrring Employing a color table thresholding segmentation technique on YCbCr to extract the visual target 3D Geometrical modeling for estimation and prediction of target pose As a future work, we are going to implement the applied algorithm on an embedded system to develop a visual RADAR System Conclusion and Future Work

55. 55