**5. Conclusions**

16 Will-be-set-by-IN-TECH

(a) (b) (c) (d)

(e) (f) (g) (h) Fig. 7. Examples of input test scenes with a geometrically ditorted versions of the target and

by 10 degrees, (c) Target with size enlarged by 20%, (d) Target with size reduced by 20%. Output correlation intensity plane obtained with AUF: (e) for scene shown in (a), (f) for scene

by the ACF on each scene. We can see one sharp correlation peak in each output intensity plane, indicating the presence of the target at the correct position. Moreover, observe that the output-correlation intensity values in the background area are very low in all the tests. Next, we compare the recognition performance of all considered composite filters when different views of the target are embedded into the background at unknown coordinates, and the

trials of each experiment for different views of the target and realizations of random noise processes were carried out. With 95% confidence the performance results in terms of DC and LE are presented in Table 1. One can observe that the proposed ACF yields the best results in terms of DC and no location errors occurred. This means that the proposed ACF is robust to

Now, we design an adaptive unconstrained filter (AUF) trained to recognize the five views of the target including rotated versions from -10 to 10 degrees with increments of two degrees, and scaled versions with 0.8 and 1.2 scale factors. In this case, the true-class training set {*T*} contains 70 training images. The AUF was synthesized using the iterative training algorithm shown in Fig. 2, reaching its maximum value in terms of the objective function "*J*(**h***AUF*)" (see Ec. (37)) after 16 iterations. The normalized performance of the AUF in the design process in terms of *J*(**h***AUF*) versus the iteration index is shown in Fig. 6. To illustrate the performance of the AUF in recognizing geometrically distorted views of the target, Fig. 7

shown in (b), (g) for scene shown in (c), (h) for scene shown in (d).

*<sup>n</sup>* = 2/256. (a) Target rotated by -10 degrees, (b) Target rotated

*<sup>n</sup>* is changed. To guarantee correct statistical results, 120 statistical

with additive noise variance *σ*<sup>2</sup>

variance of additive noise *σ*<sup>2</sup>

additive noise and to background disjoint noise.

In summary, the chapter presents an iterative approach to synthesize adaptive composite correlation filters for object recognition. The approach can be used to monotonically improve the quality of a simple composite filter in terms of quality metrics using all available information about the target object to be recognized, and false patterns to be rejected such as the background. Given a subset of true-class training images the proposed approach designs the impulse response of an optimized adaptive filter in terms of a particular performance criterion using an incremental search-based strategy. We designed an adaptive constrained filter with the suggested iterative algorithm optimizing the discrimination capability. According to the simulation results, the proposed adaptive constrained filter proved to be very robust in recognizing different views of a target within an input scene that is corrupted with additive noise. Moreover, the filter exhibits high levels of discrimination capability and location accuracy when compared with conventional MACE and MACH formulations. Furthermore, we synthesized an adaptive unconstrained composite filter optimized with respect to a proposed objective function based on the ACH, ACE, and ASM metrics. Here again, the experimental results suggest that the adaptive unconstrained filter provides a robust detection of geometrically distorted versions of the target when it is embedded within a highly cluttered background.

Finally, we can envision several lines of future research that can be derived from the algorithms and methods presented here. First, future experimental tests should consider real-world scenarios and applications to validate the usefulness of these filters in applied domains. Second, while the adaptive design process presented here has shown promising

Javidi, B. & Hormer, J. L. (1994). *Real-Time Optical Information Processing*, Academic Press, San

Advances in Adaptive Composite Filters for Object Recognition 109

Javidi, B. & Horner, J. L. (1989). Single spatial light modulator joint transform correlator, *Appl.*

Javidi, B. & Wang, J. (1994). Design of filters to detect a noisy target in nonoverlapping

Javidi, B. & Wang, J. (1997). Optimum filter for detecting a target in multiplicative noise and

Javidi, B., Zhang, G. & Parchekani, F. (1996). Minimum square error filter for detecting a noisy

Kerekes, R. A. & Vijaya-Kumar, B. V. K. (2006). Correlation filters with controlled scale

Kerekes, R. A. & Vijaya-Kumar, B. V. K. (2008). Selecting a composite correlation filter design:

Kober, V. & Campos, J. (1996). Accuracy of location measurement of a noisy target in a

Kober, V., Seong, Y. & Choi, T. (2000). Trade-off filters for optical pattern recognition with nonoverlapping target and scene noise, *Patt. Rec. Image Anal.* 10(1): 149–151. Lowe, D. G. (2004). Distinctive image features from scale-invariant keypoints, *Int. J. Comput.*

Mahalanobis, A., Vijaya-Kumar, B. V. K. & Casasent, D. (1987). Minimum average correlation

Mahalanobis, A., Vijaya-Kumar, B. V. K., Song, S., Sims, S. R. F. & Epperson, J. F. (1994).

Martinez-Diaz, S., Kober, V. & Ovseyevich, I. A. (2008). Adaptive nonlinear composite filters for pattern recognition, *Pattern Recognition and Image Analysis* 18(4): 84383. Nevel, A. V. & Mahalanobis, A. (2003). Comparative study of maximum average correlation

Nicolás, J., Campos, J., Iemmi, C., Moreno, I. & Yzuel, M. J. (2001). Convergen optical correlator alignment based on frequency filtering, *Appl. Opt.* 41: 1505–1514. Olague, G. & Trujillo, L. (2011). Evolutionary-computer-assisted design of image operators

Pérez, C. B. & Olague, G. (2008). Learning invariant region descriptor operators with genetic

Ramos-Michel, E. M. & Kober, V. (2008). Adaptive composite filters for pattern recognition in

Refregier, P. (1999). Bayesian theory for target location in noise with unknown spectral density,

Sanders, J. & Kandrot, E. (2010). *CUDA by Example: An Introduction to General-Purpose GPU*

*(ICPR 2008), December 8-11, 2008, Tampa, Florida, USA*, IEEE, pp. 1–4. Rakvic, R. N., Ngo, H., Broussard, R. P. & Ives, R. W. (2010). Comparing an FPGA to a cell for

linearly degraded and noisy scenes, *Opt. Eng.* 47(4): 82203.

that detect interest points using genetic programming, *Image Vision Comput.*

programming and the f-measure, *19th International Conference on Pattern Recognition*

an image processing application, *EURASIP Journal on Advances in Signal Processing*

background noise, *J. Opt. Soc. Am. A* 11: 2604–2612.

additive noise, *J. Opt. Soc. Am. A* 14(4): 836–844.

response, *IEEE Trans. Imag. Proc.* 15(7): 1794–1802.

a survey and comparative study, *Opt. Eng.* 47(6): 067202.

nonoverlapping background, *J. Opt. Soc. Am. A* 13(8): 1653–1666.

Unconstrained correlation filters, *Appl. Opt.* 33(17): 3751–3759.

height filter variants using ladar imagery, *Opt. Eng.* 42(541).

target in background noise, *Appl. Opt.* 35(35).

energy filters, *Appl. Opt.* 26: 3633–3640.

Diego.

*Opt.* 25: 1027–1032.

*Vision* 60(2): 91–110.

29: 484–498.

2010(764838).

*J. Opt. Soc. Am. A* 16: 276–283.

*Programming, First Edition*, Addison-Wesley.

performance, it is evident that we cannot assume that an optimal strategy has been chosen. For instance, the proposed algorithm follows an iterative mechanism to build the final solution; i.e., it incrementally constructs the training set of images. However, from a search and optimization stand-point there is no reason to assume that this is in any way an optimal strategy for the filter design process. Therefore, it would be instructive to propose, design, and test other iterative search algorithms, such as population-based meta-heuristics, since the structure of the search space is not known a priori and is probably discontinuous and highly multi-modal. Finally, a comparative study of the developed composite filters with other object recognition approaches, particularly feature based methods, might provide a more comprehensive understanding regarding the domain of competence of each.

## **6. Acknowledgements**

The research reported in this chapter was supported by Consejo Nacional de Ciencia y Tecnologia (CONACYT) and Secretaria de Investigacion y Posgrado del Instituto Politecnico Nacional (SIP-IPN), Mexico, through the projects CONACYT-0130504 and SIP20110110.

#### **7. References**


18 Will-be-set-by-IN-TECH

performance, it is evident that we cannot assume that an optimal strategy has been chosen. For instance, the proposed algorithm follows an iterative mechanism to build the final solution; i.e., it incrementally constructs the training set of images. However, from a search and optimization stand-point there is no reason to assume that this is in any way an optimal strategy for the filter design process. Therefore, it would be instructive to propose, design, and test other iterative search algorithms, such as population-based meta-heuristics, since the structure of the search space is not known a priori and is probably discontinuous and highly multi-modal. Finally, a comparative study of the developed composite filters with other object recognition approaches, particularly feature based methods, might provide a

The research reported in this chapter was supported by Consejo Nacional de Ciencia y Tecnologia (CONACYT) and Secretaria de Investigacion y Posgrado del Instituto Politecnico Nacional (SIP-IPN), Mexico, through the projects CONACYT-0130504 and SIP20110110.

Aguilar-Gonzalez, P. M., Kober, V. & Ovseyevich, I. A. (2008). Pattern recognition in

Alkanhal, M., Vijaya-Kumar, B. V. K. & Mahalanobis, A. (2000). Improving false

Bahri, Z. & Kumar, V. B. K. V. (1988). Generalized sinthetic discriminant functions, *J. Opt. Soc.*

Brown, M., Hua, G. & Winder, S. (2011). Discriminative learning of local image descriptors,

Diaz-Ramirez, V. H. (2010). Constrained composite filter for intraclass distortion invariant object recognition, *Optics and Lasers in Engineering* 48(12): 1153–1160. Diaz-Ramirez, V. H. & Kober, V. (2007). Adaptive phase-input joint transform correlator, *Appl.*

Diaz-Ramirez, V. H., Kober, V. & Alvarez-Borrego, J. (2006). Pattern recognition with an

Diaz-Ramirez, V. H., Campos-Trujillo, O., Kober, V. & Aguilar-Gonzalez, P. M. (2012).

Gonzalez-Fraga, J. A., Kober, V. & Alvarez-Borrego, J. (2006). Adaptive synthetic discriminant function filters for pattern recognition, *Opt. Eng* 45: 0570051–05700510. Goudail, F. & Refregier, P. (2004). *Statistical Image Processing Techniques for Noisy Images, an*

Hester, C. F. & Casasent, D. (1980). Multivariant technique for multiclass pattern recognition,

Multiclass pattern recognition using adaptive correlation filters with complex

adaptive joint transform correlator, *Appl. Opt.* 45(23): 5929–5941.

*application-oriented approach*, Kluwer Academic, Boston.

*IEEE Trans. Pattern Anal. Mach. Intell.* 33(1): 43–57.

nonoverlapping background with noisy target image, *Pattern Recognition and Image*

alarm capabilities of the maximum average correlation height filter, *Opt. Eng.*

more comprehensive understanding regarding the domain of competence of each.

**7. References**

**6. Acknowledgements** 

*Analysis* 20(2).

39: 1133–1141.

*Am. A* 5: 562–571.

*Opt.* 46(26): 6543–6551.

*Appl. Opt.* 19: 1758–1761.

constraints, *Opt. Eng.* 51(3):to appear.


**6**

I. Lopez-Juarez

*Mexico* 

**The Use of Contour, Shape and**

**Approach for Object Recognition** 

*Centro de Investigacion y de Estudios Avanzados del IPN (CINVESTAV)* 

How objects are recognised by humans is still an open research field. But, in general there is an agreement that humans recognise objects as established by the similarity principle – among others- of the Gestalt theory of visual perception, which states that things which share visual characteristics such as contour, shape, form, size, colour, texture, value or orientation will be seen as belonging together (Ellis, 1950). This principle applies to human operators; for instance, when an operator is given the task to pick up a specific object from a set of similar objects; the first approaching action will probably be guided solely by visual information clues such as shape similarity. But, if further information is given (i.e. type of

The task described above can also be accomplished by automated systems such as industrial robots that can be benefited from the integration of a robust invariant object recognition capability following the above assumptions and by using image features from the object's contour (boundary object information), its shape (i.e. type of curvature or topographical surface information) and form (depth information). These features can be concatenated in order to form an invariant vector descriptor which can be mapped into specific objects using Artificial Intelligence schemes such as Artificial Neural Network (ANN). In previous work, it was demonstrated the feasibility of the approach to learn and recognise multiple 3D working pieces using its contour from 2D images and using a vector descriptor called the Boundary Object Function (BOF) (Peña-Cabrera, et al., 2005). The BOF exhibited invariance with different geometrical pieces, but did not consider surface topographical information. In order to overcome this condition and to have a more robust descriptor, a methodology that includes a shape index using the Shape From Shading (SFS) method (Horn, 1970) is presented as well as the depth information coming from a stereo vision system. The main idea of the approach is to concatenate three vectors, (BOF+SFS+DI) so that not only the contour but also the object's curvature information (shape) and form are taken into account

In this article after presenting related work in Section 2 and original work in Section 3, the contour vector description (BOF), the SFS vector and the stereo disparity map (Depth) are explained in Sections 4, 5 and 6 respectively. A description of the learning algorithm using

surface), then a finer clustering could be accomplished to identify the target object.

**1. Introduction** 

by the ANN.

**Form in an Integrated Neural**

