Comparison of compression Drowsiness Detection tools in H.

264 and VC-1

Rise an Alarm.  .  Steps Involved in our method:  Step1: Continuously Monitoring Driver’s Eye to check if it “open” or “Close” Steo2: If Eyes are Closed Continuously for some xxx seconds.Introduction  Detecting Drowsiness of the Driver by finding his state of “eye” .Open or Closed.

. .Introduction cont.

.  Method 2: Using Morphology and other Image Processing Techniques..Introduction cont.  We did step1(Eye State Detection) in two methods  Method 1: Using Opencv’s Haar Detect Objects.

Method 1: Eye State Detection using OpencV Approach Followed: Input frame Face Detection No Face Detected ? Skip the frame Set ROI to a portion of face region • Discard Few Pixels from top(Skip Forehead) • Process only 1/3 rd of face Height Detect for Eye in ROI Eye Detected ? No Eye State = OPEN Yes Eye State = CLOSE .

a set of features distinctive features are extracted that can be used to classify the object(Eye) are selected Using the Generated Classifier (with a set of features that give that chance to find the desired object) for Object(Eye) Detection in test input/image .Eye Detection using OpencV Haar Detect Objects  How OpenCV Eye Detection Works Training and Generating a Classifier using Haar Features.

a Classifier is trained. that are scaled to the same size (say. CASCADED and BOOSTED. . Car. Ear..Training a Haar Classifier  Initially proposed by Paul Viola and improved by Rainer Lienhart.  Negative examples .  A few thousands of sample views of a particular object (Eye).  With those Positive and Negative Samples given. lips.arbitrary images(that are not an eye: Nose. chair etc. 15x10) are to be collected.) of the same size are also required. called positive examples.

Detection Using the Trained Classifier The process of Eye detection slides a “search window” through the image. but since Eye in an image can be different from the assumed scale. searching for the object across a ranges of sizes. checking whether an image region can be considered as a “Eye object” or not. Search Window E1 E2 Not an Eye En Eye . the “search window” goes trough the image several times. The detector assumes a fixed scale for the object.

weights. thresholds etc. . no of stages.Code point View of Trained Haar Classifier  Opencv expects a trained classifier in the “xml” format containing all the information : Minimum window size..

/* the source image. 15) /* minimum detection scale */ ). eyes = cvHaarDetectObjects( img.e..5). storage. /* scale factor.0 ) ). cvRect( r->x. /* the eye classifier */ /* memory buffer */ 1. Image with Face Detected */ /* x = start from leftmost */ r->y + (r->height/5. 0. /* y = a few pixels from the top */ r->width. /* the source image i.2. . r->height/3. neighborhood. do_canny_purring*/ cvSize(15.Code: Haar Detect Objects cvSetImageROI(img. 4. with the estimated location defined */ /* width = same width with the face */ /* height = 1/3 of face height */ cascade_eye.

Using OpenCV’s function for Embedded      Extracting the Required functions Tuning according to the requirement of Embedded Floating Point to Fixed Point Conversion Generating Classifier for as per Our Requirement Turning Haar Trained Classifiers which are in the form of xml files into constant arrays. .

Before Optimization : 110 – 130mS After Optimization : 50 – 60 mS . Optimized and Validated the same to IPNC’s floating point DSP C674x+.  @ 500 MHz .Porting and Integration  We ported. processing time for eye detection. doing all the required steps.

Advantages  Detects the Eye very Efficiently  Works good in high and medium light conditions  Less False Detection .

rising false alarms (Can be overcome by Training a classifier with a positives having wide range of spectacles ).  Allows tilting of face upto some angle only.Disadvantages  Less or No Detection in very poor Light / Night (Can be overcome by using high Range powerful IR LEDS along with Cam). . hw engine is not having this issue). (If face is tilted greater than around 45degrees. it can’t able to detect eyes – problem with our FD only.  If a person wears very powerful Anti-glare spectacles and some kind of very complex framed spectacles eye detection rate will be less.

Sign up to vote on this title
UsefulNot useful