英语翻译3.1 TrackingMore often than not,the false positive detec

英语翻译
3.1 Tracking
More often than not,the false positive detections from our license plate detector were erratic,and if on the car body,their position was not temporally consistent.We use this fact to our advantage by tracking candidate license plate regions over as many frames as possible.Then,only those regions with a smooth trajectory are deemed valid.The tracking of license plates also yields a sequence of samplings of the license plate,which can be used as input to a super resolution preprocessing step before OCR is performed on them.
Numerous tracking algorithms exist that could be applied to our problem.Perhaps the most well-known and popular is the Kanade-Lucas-Tomasi (KLT) tracker.The KLT tracker makes use of a Harris corner detector to detect good features to track in a region of interest (our license plate) and measures the similarity of every frame to
the first allowing for an affine transformation.Sullivan etal.make use of a still camera for the purposes of tracking vehicles by defining regions of interest (ROI) chosen to span individual lanes.They initiate tracking when a certain edge characteristic is observed in the ROI and make predictions on future positions of vehicles.Those tracks with a majority of accurate predictions are deemed valid.Okumaet al.use the Viola and Jones framework to detect hockey players and then apply a mixture particle filter using the detections as hypotheses to keep track of the players.
Although each of these tracking methods would probably have worked well in our application,we chose a far simpler approach which worked well in practice.Because detecting license plates is efficient we simply run our detector on each frame and for each detected plate we determine whether that detection is a new plate or an instance of a plate already being tracked.To determine whether a detected plate is new or not,the following conditions are checked:
• the plate is within T pixels of an existing tracker
• the plate is within T` pixels of an existing tracker and the plate is within θ degrees of the general direction of motion of the plates in the tracker’s history
If any of these are true,the plate is added to the corresponding tracker,otherwise a new tracker is created for that plate.In our application T` was an order of magnitude larger than T.Our tracking algorithm was also useful for discarding false positives from the license plate detector.The erratic motion of erroneous detections usually resulted in the initiation of several trackers each of which stored few image sequences.Image sequences of 5 frames or fewer were discarded.
3.2 Character Recognition
It was our initial intent to apply a binarization algorithm,such as a modified version of Niblack’s algorithm as used by Chen and Yuille,on the extracted license plate images from our detector,and then use the binarized image as input to a commercial OCR package.We found,however,that even at a resolution of 104 × 31 the OCR packages we experimented with yielded very poor results.Perhaps this
should not come as a surprise considering the many custom OCR solutions used in existing LPR systems.
谢绝直接使用工具而读起来不通的！