Técnicas de reconocimiento de objetos en aplicaciones reales

Resum

This thesis evaluates and proposes object description and retrieval techniques in different real applications. It addresses the classification of boar spermatozoa according to the acrosome integrity using several proposals based on invariant local features. In addition, it provides two new methods for inserts localisation and an automatic solution for the recognition of broken inserts in edge profile milling heads that can be set up on-line without delaying any machining operations. And finally, it evaluates different keypoints clustering configurations for object retrieval and proposes a new descriptor, named colour COSFIRE, in the scope of the Advisory System Against Sexual Exploitation of Children project. Automatic assessment of sperm quality is an important challenge in the veterinary field. In this dissertation, we studied the description of boar spermatozoa acrosomes using image analysis to automatically classify them as intact or damaged. We characterised the acrosomes using invariant local features, particularly SIFT and SURF, improving the results obtained with global texture descriptors. The best results were achieved for the classification of SURF descriptors by k-NN. The overall accuracy was 94.88%, with a higher hit rate in the damaged class, 96.86%, than in the intact one, 92.89%. The opposite behaviour, higher hit rate in the intact class, was yielded by global texture descriptors. In order to overcome the classification of invariant local features with a support vector machines (SVM), we presented an approach which successfully deals with having more than one descriptor per image. Interest points were detected and described using SURF. Our method classifies spermatozoa heads, exploiting that a head usually contains more distinctive points of their own class than doubtful points which could be misclassified. Experiments showed an accuracy of 90.91% (94.94% and 86.87% for the intact and damage classes respectively) which indicates that this approach could be an alternative to consider for classifying invariant local features descriptors. We also proposed an early fusion of invariant local features with global texture descriptors to study the integrity of the head acrosomes, evaluating both SVM with bag of visual words (BoW) and k-NN for the classification. The concatenation of SURF with Legendre descriptors achieved an accuracy of 95.56% (93.63% in the intact and 97.48% in the damaged class) when classifying using kNN, outperforming the results obtained for both descriptors separately. Wear evaluation of inserts is a key issue for extending lifetime of cutting tools and ensuring high quality of products. In this thesis, we introduced two image processing methods to automatically localise cutting tools in an edge profile milling head and another one to determine if they are broken. Unlike other machining operations presented in the literature, we were dealing with edge milling head tools for aggressive machining of thick plates (up to 12 centimetres) in a single pass. The studied cutting head tool is characterized by its relatively high number of cutting tools (up to 30) which makes the localisation of inserts a key aspect. We detected the screws that fasten the inserts using a circular Hough transform. In a cropped area surrounding a detected screw, we used Canny’s algorithm and a standard Hough transform to localise line segments that characterise insert edges. Considering this information and the geometry of the insert, we identified which of these line segments is the cutting edge. The output of our algorithm is a set of quadrilateral regions around the identified cutting edges that can be used as input to other methods specialised in assessing the state of the cutting edge. Our proposal is very effective (accuracy equals to 99.61%) for the localisation of the cutting edges of inserts in an edge profile milling machine. Following up this result, we studied how to recognise broken inserts because it is critical for a proper tool monitoring system. The method that we presented first localises the screws of the inserts and then determines the expected positions and orientations of the cutting edges using known geometrical information. We computed the distances, called deviations, between the expected cutting edge and the real one to determine if it is broken. We evaluated the proposed method on a new dataset that we created and made publicly available. The obtained results, with a harmonic mean of precision and recall equals to 91.43%, showed that this algorithm is effective and suitable for the recognition of broken inserts in machining head tools. Finally, we proposed a more generic and versatile approach for the localisation of inserts based on trainable COSFIRE filters. It can be automatically configured regardless of the appearance of the inserts. A new function for the computation of the response of the COSFIRE filter was also introduced, outperforming the previous ones. Results, with a harmonic mean of precision and recall equals to 89.89%, improved preceding works based on template matching. Altogether, the results obtained for this application foster further implementation at a working manufacturing environment. Advisory System Against Sexual Exploitation of Children European project aims to provide a technological solution to help the fight against child pornography. One of the most challenging tasks in this project was the retrieval of specific objects from collections with a huge amount of images and videos. We evaluated different clustering configurations of SIFT keypoints in relation with their pose parameters: coordinates location, scale and orientation. On the one hand, we used the similarity measure of the closest pairs of keypoint descriptors. On the other hand, we used a Hough transform, with different parametrization values, to identify clusters of at least three points voting for the same pose of an object and we verified the consistency of the pose parameters with the least squares algorithm. Results were computed for a publicly available dataset of 614 images illustrating possible sceneries of a real case. Higher precisions were obtained without clustering at small cuts of the hit list, whereas better precisions were yielded with Lowe’s clustering at high cuts. Moreover, colour COSFIRE filters were proposed for the retrieval of colour objects. They add colour description and discrimination power to COSFIRE filters as well as provide invariance to background intensity. Colour COSFIRE filters were presented both for patterns made up of colour lines and for patterns that are colour objects, outperforming standard COSFIRE filters both for retrieval and classification tasks. The work proposed in this thesis contributes to the understanding and resolution of real applications using object recognition and image classification techniques.