Proximity Sensor Detection

The concept of context awareness and micro-environment sensing can be used to develop many applications based on inbuilt sensors which will be able to simulate the higher level applications of Smartphones. Automatic call acceptance It comes under phone interaction detection category. There are some situations in our daily routine when we are not able to pick phone because we need to swipe to pick up a call e.g. stuck in traffic, at railway station, markets etc. In such situations it is possible to pick the call automatically with the help of position of phone with respect to user using proximity sensor. Proximity sensor detects the object in range of 2 to 5 cm. Closed environment settings It comes under the local placement detection category.

Although the front-mounted proximity sensor can perceive sheltering in front, the phone is unaware of that backwards. Thus with proximity sensor alone, it is likely to miss some ‘in-hand’ cases, e.g., when the user is making a phone call with his ear covers the front end of the phone. Therefore we also employ the back mounted camera for proximity perception backwards. The rationale is that the global contrast of a photo taken in a closed environment (e.g., in-pocket) is usually low, which is reflected in the gray-scale histogram of the photo. As a motivating experiment, we collect photos taken by a background photographing application for various phone placements in diverse scenarios, including chest pocket, pants, bags and hands in supermarkets, cafes and streets. demonstrates the gray-scale histogram distribution of photos in six different conditions. The upper three correspond to closed environments including in bags, chest pockets and pants, while the lower three are in-hand situations. In general, the histogram distributes more spread-out when the phone is held in hand than placed in closed environments, indicating higher global contrast due to better lighting conditions. To quantitatively measure the extent of dispersion of the gray-scale histogram, we calculate the average slope of gray-scale pixels between two quantiles q1 and q2 in its CDF. Fig. 3.2 plots the CDF for the six situations. Fig 3.2: CDF of grey-scale histograms under various phone