Car detection in real time environment during fog hazy day
Images taken in poor weather conditions have a high amount of haze (fog) in it. It degrades the visibility and clarity of the image. In this project, we propose the technique to enhance the haze image and to detect the object present in it. The dehazing technique used here is multi region segmentation, multi region fusion method and exposure fusion method. Finally, the enhanced image is segmented based on Saliency Map and then vehicles are detected using thresholding and morphological operations. As per the image dehazing and vehicle or car detection, our proposed method can alert the driver for prevent the accident. An overall accuracy of almost maximum is achieved regarding image dehazing and vehicle detection of our proposed method.
PROJECT OUTPUT VIDEO:
- To enhance the haze (fog) images.
- To detect the object using Saliency map.
- In existing, they proposed murk/haze removal by using CLAHE and object detection using Curvelet Transform.
- Initially the RGB image is converted to gray scale image.
- The gray scale image is then filtered using median filter.
- The filtered image is enhanced using CLAHE.
- Finally the enhanced image is segmented by Curvelet Transform.
DISADVANTAGES OF EXISTING SYSTEM:
- In existing less amount of haze/murk removal in image is proposed.
- The rate of detection accuracy was less.
The proposed model follows multi region segmentation, multi region fusion method, exposure fusion method for dehazing and saliency map for car detection. It consists of five steps.
Step 1: Collecting the test images from public database.
Step 2: Estimate the global atmospheric light based on multi region segmentation.
Step 3: A new multi region fusion method is used to optimize the transmission.
Step 4: An exposure fusion method is constructed to dehaze and improve the image quality
Step 5: The dehazed image is segmented and then cars are detected based on saliency map, thresholding and morphological operations.
Step 6: To avoid the accident, some alert system provided to driver.
ADVANTAGES OF PROPOSED SYSTEM:
- The high level of performance occurs in removal of haze.
- The rate of detection accuracy is high.
- System : Pentium Dual Core.
- Hard Disk : 120 GB.
- Monitor : 15’’ LED
- Input Devices : Keyboard, Mouse
- Ram : 2 GB
- Operating system : Windows 10.
- Coding Language :
- Tool : MATLAB R2013A /2018